[Raleigh Martin] 13:01:29
On detail in our new division of research, innovation, synergies, and education, or rise for sure.

[Raleigh Martin] 13:01:35
In the directorate for geosciences i am joined with several colleagues here today from the geoscience directorate including the earth science division office of polar programs atmospheric and geospace science and ocean sciences who are listed on the screen here.

[Raleigh Martin] 13:01:54
I'm also joined by colleagues from our computer and information systems and engineering director at our size, which is a co-sponsor of this program.

[Raleigh Martin] 13:02:04
This includes folks from our Office of Advanced cyber infrastructure and division of intelligent information systems. And then finally, a third co-sponsor on this solicitation is the directorate for Mathematical and Physical Sciences Division of Mathematical Sciences.

[Raleigh Martin] 13:02:21
So, so joined by colleagues from there as well. Today I'm going to give a presentation on our new program solicitation, collaborations in AI and geosciences are CAI for short.

[Raleigh Martin] 13:02:33
This is program solicitation NSF, 24 dash 5 1 8. With proposals do. No later than March, fifteenth, 2024.

[Raleigh Martin] 13:02:46
So a few logistics before we get into the detailed information of the webinar. First, I wanted to note that.

[Raleigh Martin] 13:02:54
We are going to start with about a 2025 min presentation followed by QA. The entirety of this webinar will be recorded.

[Raleigh Martin] 13:03:03
And it will be posted on the CAIG program page at this link. Following this presentation, as I said, there will be an opportunity for Q&A, which I'll talk about in the next slide.

[Raleigh Martin] 13:03:15
I will note that in general, the, the Q&A today will be focused on general questions about the solicitation.

[Raleigh Martin] 13:03:21
If you have more specific questions about your specific proposal idea, we would encourage you to reach out to one of the program officers listed on the CAIG program page.

[Raleigh Martin] 13:03:35
So another housekeeping note. This is a Zoom Webinar. The only folks you can see on the screen are NSF staff who are giving this presentation answering questions.

[Raleigh Martin] 13:03:49
All their attendees are muted and webcams are disabled. If you would like to see a live closed captioning, please click the live transcript button in your, Zoom Webinar, module.

[Raleigh Martin] 13:04:00
And if you want to ask a question at any time, please feel free to do so using the Q&A feature in Zoom.

[Raleigh Martin] 13:04:07
You can send in the question whenever you want and we'll start answering those. When the the presentation portion of the webinar is done.

[Raleigh Martin] 13:04:14
You may choose by clicking on the send anonymously button to not reveal your identity when asking the question that's up to you.

[Raleigh Martin] 13:04:23
Okay, so. With that housekeeping done, I'm now going to move into the kind of substantive part of the webinar.

[Raleigh Martin] 13:04:31
The first part of the presentation I'm going to give an overview of the CAI G program.

[Raleigh Martin] 13:04:36
Why we created this program. What some of the priorities are for the program. The scope of what is and is not supported.

[Raleigh Martin] 13:04:44
And in relation to other programs that may be of interest for AI driven research or geoscience related research at NSF.

[Raleigh Martin] 13:04:54
And I'm also gonna talk a bit about some related initiatives for the National Discovery Cloud for Climate and DCC.

[Raleigh Martin] 13:04:59
And the national AI research resource or near.

[Raleigh Martin] 13:05:07
Okay, so. Motivation first. I think it's important to begin with. The working definition that we are using in the solicitation for artificial intelligence, we refer to this as intelligence that comes from any computational tool.

[Raleigh Martin] 13:05:23
Including logic and decision trees. AI includes approaches like machine learning methods that use optimization and statistical techniques to enable machines to perform tasks.

[Raleigh Martin] 13:05:34
And make discoveries after exposure to data. We recognize that the definition of AI is evolving. And may be context specific.

[Raleigh Martin] 13:05:43
The, purpose of providing this definition is to simply get folks on the same page in terms of what it is that we're generally seeking to do in the solicitation.

[Raleigh Martin] 13:05:52
So in terms of motivation and why we have an AI focused program for the geosciences. A big motivator here is that we've been observing a market increase in usage of innovative AI driven approaches in the geosciences in recent years.

[Raleigh Martin] 13:06:08
Increasingly, these approaches are not simply taking AI approaches off the shelf. But they're actually, providing their own novel, innovation or kind of novel integration of AI driven methods for geoscience questions.

[Raleigh Martin] 13:06:23
This was recognized, for example, in the recent National Academies report on next generation Earth system sciences for NSF.

[Raleigh Martin] 13:06:31
We also realized that utilization of AI driven approaches requires building both technical and human capacity for the adoption of those approaches.

[Raleigh Martin] 13:06:42
This is a goal. That is going to be addressed through the new National AI Research Resource Activity, which I'll talk about a little bit later.

[Raleigh Martin] 13:06:56
I research resource activity, which I'll talk about a little bit later. I'm going to be addressed through the new National AI research resource activity, which I'll talk about a little bit later.

[Raleigh Martin] 13:07:10
But more generally, and in this solicitation, we are also

[Raleigh Martin] 13:07:11
So with that in mind, there are 3 primary priorities for the CAG solicitation. First, to enable significant breakthroughs in geosciences research questions through advancement and AI techniques.

[Raleigh Martin] 13:07:20
And our innovative uses of sophisticated AI methods that overcome. Geoscience research challenges or bottlenecks.

[Raleigh Martin] 13:07:28
I wanted to emphasize that last point. Using AI techniques to address problems that are already, you know, well handled and we already have existing methods.

[Raleigh Martin] 13:07:38
It's not necessarily gonna cut it in the solicitation. It really needs to be driven by something that could not otherwise be addressed.

[Raleigh Martin] 13:07:44
But for the adoption of innovative AI techniques to overcome that challenge or bottleneck. A second priority is for the solicitation is to expand access to and usability of geoscience data and existing cyber infrastructure or CI for sure, including training data sets and benchmarking to enable innovative AI driven approaches in the geosciences.

[Raleigh Martin] 13:08:07
This includes what is not limited to efforts to federate and provide access to cyber infrastructure for research on the intersection of AI and climate science.

[Raleigh Martin] 13:08:15
So here we're talking about, the basically the lowering the technical barriers to using AI driven approaches in the geosciences.

[Raleigh Martin] 13:08:27
The third priority is to build a workforce capacity for using advanced AI methods in the geosciences through educational and broadening participation activities.

[Raleigh Martin] 13:08:34
With an emphasis on inclusion of historically underrepresented groups in STEM and underserved institutions.

[Raleigh Martin] 13:08:41
Such as non R one minority serving institutions, two-year colleges. Primarily undergraduate institutions.

[Raleigh Martin] 13:08:47
And emerging research and masters level institutions. So for the purpose of this solicitation, we encourage these activities that address geoscience workforce development across a range of career stages to increase expertise on AI methods.

[Raleigh Martin] 13:09:03
And contribute to geoscience research discoveries. We acknowledge that this upskilling needs to occur across all career stages.

[Raleigh Martin] 13:09:11
And we note that these types of capacity building activities are supported within the solicitation, but should be pursued in tandem with activities for AI innovation and or accessibility.

[Raleigh Martin] 13:09:21
In the geos is basically the first 2 priorities of the solicitation.

[Raleigh Martin] 13:09:30
Okay, so another thing to make sure is clear here. Is that this is a collaborations and AI for the geosciences.

[Raleigh Martin] 13:09:39
So we want to make sure that folks are on the same page. By what we mean by geosciences.

[Raleigh Martin] 13:09:44
Specifically in the solicitation we are referring to those academic research communities that are supported by our geoscience directorate at NSF.

[Raleigh Martin] 13:09:51
Which includes atmospheric and geospace sciences, ocean sciences, earth sciences, and polar programs, which are represented by our 4 disciplinary offices and divisions in the geoscience directorate.

[Raleigh Martin] 13:10:06
I'll also note that this program is being led by our new division of research innovation synergies and education rise.

[Raleigh Martin] 13:10:14
Which encompasses trans-disciplinary and societal impact activities within the director of geosciences.

[Raleigh Martin] 13:10:22
So to understand the disciplinary focus of your proposal or of what is accommodated in this program, I would encourage you to look at program pages in the geosciences to see what is and is not in the geoscience as we define it here.

[Raleigh Martin] 13:10:38
Okay, so. Want to make clear what is and is not supported by the the CAIG solicitation.

[Raleigh Martin] 13:10:47
First we are really looking for some level of innovation in AI driven approaches, whether that's through development of new approaches or through a novel integration of existing approaches.

[Raleigh Martin] 13:10:59
If instead your goal is to Take existing AI driven methods. And apply them to geoscience questions.

[Raleigh Martin] 13:11:08
So the focus is really on the geoscience question, but there's not really the innovation in AI.

[Raleigh Martin] 13:11:12
That's fine, but not for this program. That's better suited to. Our disciplinary science programs in the geoscience directorate.

[Raleigh Martin] 13:11:19
If your goal is to primarily focus on innovative AI driven research. Perhaps there are geoscience applications or data sets that are being used, but these are really just use cases.

[Raleigh Martin] 13:11:31
There's not a geoscience. Bottleneck or research question that is being overcome. That there are programs for that at NSF, probably not the CAI G program.

[Raleigh Martin] 13:11:40
Would encourage you to look at programs within our. Size division of intelligent information systems, IS and the NPS Division of Mathematical Sciences DMS.

[Raleigh Martin] 13:11:51
If the focus is primarily on the development of cyber infrastructure that enables AI driven research, That's also very interesting, probably not the focus of this solicitation but better suited to programs in our Office of Advanced cyber infrastructure such as CSSI cyber infrastructure for sustained scientific innovation.

[Raleigh Martin] 13:12:09
If the focus is on improving cyber infrastructure literacy or skills development. Again, an interesting topic maybe outside of CAIG encourage you to look at programs like cyber training.

[Raleigh Martin] 13:12:21
To learn more. If you go to this QR code or the GOCI opportunities web page, you can find out more information about a lot of these programs.

[Raleigh Martin] 13:12:33
Finally, I wanted to just briefly cover some related opportunities. This isn't directly at the purview of the CAIG program, but these are an important developments at NSF with which you may want to be aware.

[Raleigh Martin] 13:12:46
The first is that NSF recently announced its intent to support what is known as a national discovery cloud for climate.

[Raleigh Martin] 13:12:53
Basically, pilot projects that develop an integrated NASA scale cyber infrastructure to support end-to-end climate research and education.

[Raleigh Martin] 13:13:01
We recognize that this NDCC initiative has some intersections with the goals of CAIG as it relates to certain climate related AI driven research in the geosciences.

[Raleigh Martin] 13:13:15
If that is the case, it would encourage you to look at this to your colleague letter NSF, 24 dash 0 2 4.

[Raleigh Martin] 13:13:21
If you see an intersection with the goals of the NDCC. as described in the CCL you can indicate that on your CAI G proposal CHE is not formally listed in the NDCC, dear colleague letter, but that was just a function of timing of when the respective opportunities came out.

[Raleigh Martin] 13:13:42
So that's why I'm announcing this here today.

[Raleigh Martin] 13:13:45
Another major initiative at NSF right now is called the National AI Research Resource or Nair. You may have heard that the White House put out an executive order on AI on October thirtieth of last year.

[Raleigh Martin] 13:13:57
Expected within 90 days of that executive order or basically within about a week. Is an announcement from NSF regarding a pilot implementation of this an air, which will be an infrastructure, governance mechanisms and user interfaces to pilot.

[Raleigh Martin] 13:14:15
Initial integration of distributed computational data model and training resources to support AI related research and development. So in general, we encourage proposers to utilize advanced computing, shared advanced computing resources that are made available by NSF.

[Raleigh Martin] 13:14:32
And in particular, when, the Naira pilot is announced, we would encourage you to take a look at that to see how.

[Raleigh Martin] 13:14:38
It might support the proposal you're putting forward to this program.

[Raleigh Martin] 13:14:45
Okay, so that kind of covers a general overview of the the goals and priorities of CAIG and what we're covering.

[Raleigh Martin] 13:14:51
Now I'm gonna shift gears a bit and talk more. Specifically about some of the the provisions for proposal peppers.

[Raleigh Martin] 13:14:58
I wanted to note in general that we very, very strongly encourage you to look carefully at the entirety of the CAIG solicitation to make sure that your proposal is dressing.

[Raleigh Martin] 13:15:11
The specific requirements of the solicitation that you're aligned with the goals of the program. We also encourage you in general to always look at the proposal and award policies and procedures guide or PAP G, which provide the general guidance on NSF proposal submissions.

[Raleigh Martin] 13:15:26
And we really encourage you. Early in the process of preparing your proposal to reach out to program officers to make sure that what you're doing is within scope of the program.

[Raleigh Martin] 13:15:36
We don't want to see proposals get needlessly rejected simply because they were way outside of the scope of what we were intending to support in this program.

[Raleigh Martin] 13:15:46
We really do encourage that comp the communication early and often.

[Raleigh Martin] 13:15:52
So first, just some general things to be aware of when preparing your proposal. This is all in the CAIG solicitation, but I want to repeat it here.

[Raleigh Martin] 13:16:00
First, we ask that the title of your proposal be proceeded with the CAIG. Acronym.

[Raleigh Martin] 13:16:07
Second, we ask that in your project summary, if you're availing yourself of these resources for shared computing, which I'm going to talk about in later slides.

[Raleigh Martin] 13:16:16
For high throughput computing or HTC access for, cloud bank or cloud access. Or if you're identifying a nexus with the NDCC or National Discovery Cloud for climate.

[Raleigh Martin] 13:16:27
Please include those keywords in your project summary. For the project description. As I will describe in a subsequent slide, there are 3 specific requirements for the solicitation.

[Raleigh Martin] 13:16:39
You must include separate sections within the project description that address these 3 specific requirements, which I'll describe in a future slide.

[Raleigh Martin] 13:16:47
Finally, in terms of budget, we get a lot of questions about this. There's no specific upper limit on budget, but there are a few things to keep in mind.

[Raleigh Martin] 13:16:55
For the projects that we're going to be seeing proposals for. First, the upper limit duration we're allowing is up to 3 years.

[Raleigh Martin] 13:17:03
Second, overall, we expect to support 10 to 15 projects within anticipated 6 to 10 million dollars total.

[Raleigh Martin] 13:17:11
To be distributed across all the project. So. Again, there's not a specific upper limit. But this kind of should give a guideline to roughly the types of proposal sizes we're expecting to see in terms of budget.

[Raleigh Martin] 13:17:23
Finally, in terms of the size of the proposing team, We expect teams to be interdisciplinary.

[Raleigh Martin] 13:17:29
Including both geoscientists and experts in AI with 2 to 3 lead collaborating senior personnel.

[Raleigh Martin] 13:17:35
This includes Pi's and Ko Pi's. And associated students postdocs research software engineers and or other staff.

[Raleigh Martin] 13:17:44
So in terms of team size, that should also give you a sense of where we expect to see typical budgets.

[Raleigh Martin] 13:17:51
The, eligibility for submissions of CAIG is somewhat limited. Eligible submitting organizations may include institutions of higher education or universities.

[Raleigh Martin] 13:18:02
Non-profit, non-academic organizations. And then. Federally funded.

[Raleigh Martin] 13:18:09
Research and development centers, FFR DCs. We note in particular if you are submitting from an FFR DC you must contact us in advance.

[Raleigh Martin] 13:18:19
And we will look carefully to make sure that what you're proposing is not duplicating. What the FFR DC is or will expected to be supported to do at some point in the future.

[Raleigh Martin] 13:18:31
Other organizations not listed here such as for profits may still be involved. Through mechanisms including unfunded collaboration subaward or consulting arrangements they just cannot be the lead proposing organization.

[Raleigh Martin] 13:18:44
We also note there are some limits on senior personnel, which includes but is not limited to PI and KoPI.

[Raleigh Martin] 13:18:51
A person may be listed as senior personnel on at most 2 proposals. If we get an access number of proposals from an individual.

[Raleigh Martin] 13:19:00
Those proposals will be returned without review. Based on the order in which the proposals were received. So the third proposal that has received from a person listed as senior personnel will automatically be returned without review.

[Raleigh Martin] 13:19:12
And you can see that the definition of senior personnel in the proposal award policies and procedures guide PAPchi exhibit 2 dash 3.

[Raleigh Martin] 13:19:21
Okay, so this is very important. Proposals we received through CAAG will be reviewed through the NSF merit review process.

[Raleigh Martin] 13:19:29
This includes the 2 standard NSF merit review criteria for broader impacts, the potential to benefit society. And intellectual merit, the potential to advance knowledge.

[Raleigh Martin] 13:19:41
In addition, we have 3 additional solicitation specific criteria. Against which proposals will also be evaluated. And these align with the 3 specific requirements that are listed in the program description for CAI G.

[Raleigh Martin] 13:19:55
So the first requirement. Is geoscience advancement. In particular, we're going to be looking at how do the proposed activities contribute to advancing geoscience research and or education in response to motivating geo motivating science drivers.

[Raleigh Martin] 13:20:10
This is where we're really looking at. What is the bottleneck in geosciences research and or education and how is the use of AI driven methods overcoming that bottleneck?

[Raleigh Martin] 13:20:20
Second AI impact. This is looking at the novel integration of AI. That is being proposed and again is it overcoming significant methodological and our capacity bottlenecks that inhibit research progress in the geosciences.

[Raleigh Martin] 13:20:33
So. You know, how is this AI approach innovative? We're not looking simply for using AI methods off the shelf.

[Raleigh Martin] 13:20:40
They has to be. At the very minimum, a novel way in which the AI is being used. And or also you know some innovation in the development of those techniques and tandem with addressing that Goscience research need.

[Raleigh Martin] 13:20:53
And then finally, partnerships. We want to see meaningful interdisciplinary partnerships that support the geoscience advancement and AI impact that's described here.

[Raleigh Martin] 13:21:04
And we want to see that these partnerships are benefiting all participants. It's not just a one-way flow of information, but the geoscientists are benefiting.

[Raleigh Martin] 13:21:13
The AI experts are benefiting. And what are those pathways, for example, for cross-training of students and other researchers and the methods that are being pursued.

[Raleigh Martin] 13:21:22
Something else I wanted to note here, this solicitation, we don't really have the bandwidth support substantial development of cyber infrastructure or acquisition of hardware.

[Raleigh Martin] 13:21:33
Instead, proposers are encouraged to utilize existing cyber infrastructure resources. I would point you primarily to our access advanced cyber infrastructure coordination ecosystem services support program.

[Raleigh Martin] 13:21:49
This is the successor to the 6 exceed program. Nsf funded researchers and they request allocations for various computing shared computing resources through this access resource.

[Raleigh Martin] 13:22:01
I very strongly encourage you to look into this and consider this for your proposal. We also have specific opportunities for high throughput computing through the partnership to advance throughput computing project.

[Raleigh Martin] 13:22:10
You can request these HTC resources, via supplementary document in your proposal. If you choose to do so.

[Raleigh Martin] 13:22:17
We also have. A program for access to commercial cloud resources through the cloud bank cloud access program.

[Raleigh Martin] 13:22:25
Again, this is something you can request. Optionally via supplemental document in your proposal.

[Raleigh Martin] 13:22:32
So there are no. Supplementary documents that are required in your proposal other than those. Required.

[Raleigh Martin] 13:22:41
As standard in the proposal award policies and procedures guide Patchy. But optionally, I did one highlight a few of the supplementary documents you could be including.

[Raleigh Martin] 13:22:50
The first is this supplementary document on request for high throughput computing resources. If you choose to do so.

[Raleigh Martin] 13:22:58
Second, if you choose to request cloud credits through cloudbank. We ask that you include this supplementary document.

[Raleigh Martin] 13:23:07
One note is that there is a cost associated with those commercial cloud credits that cost should be included with those commercial cloud credits.

[Raleigh Martin] 13:23:17
That cost should be included in the supplementary document for cloud. That cost should be included in the supplementary document for Cloud Bank.

[Raleigh Martin] 13:23:18
But it should not be included on the main proposal budget page. The form 1030 a budget that kind of standard overall budget for the proposal.

[Raleigh Martin] 13:23:28
So please keep that in mind. If you have collaborative arrangements with unfunded collaborators, you can describe that in your project description.

[Raleigh Martin] 13:23:35
In your facilities, equipment and other resources section of the proposal. And then to demonstrate those the existence of those collaborative arrangements, there's kind of a standard format for a letter of collaboration letter from those unfunded collaborators which can be included as a supplementary document.

[Raleigh Martin] 13:23:52
Okay, so I've covered most of the proposal preparation information now. I just wanted to describe a few other considerations for the CAI-G program here.

[Raleigh Martin] 13:24:02
So first, as I noted in the opening, this is the initial inaugural.

[Raleigh Martin] 13:24:09
Caig competition. This is the first time we're doing this. That said, we don't have any definite plans yet for a new, CAIG competition after the, 2,024.

[Raleigh Martin] 13:24:22
Nargo competition. We expect to look at the outcomes of this initial competition, including early results from funded projects to inform whether there will be a future funding opportunity for CAIG.

[Raleigh Martin] 13:24:34
But there's no definite plan at this time. That is to be determined. In general, we do, aim to be responsive to the, the needs of the research community and directions and priorities that are expressed.

[Raleigh Martin] 13:24:47
Among the broader community of scientists.

[Raleigh Martin] 13:24:53
So I wanted to. Reiterate here, read the solicitation very carefully when preparing your proposal.

[Raleigh Martin] 13:25:02
I will note again that this presentation and the slides will be posted probably in about a week or so on the CAI program page so you can access the information there.

[Raleigh Martin] 13:25:12
And we really, really strongly encourage you to reach out to any of the program contacts listed on the solicitation.

[Raleigh Martin] 13:25:20
To reach out with further specific questions. Especially regarding the potential fit of your proposal to this call. We really do want to see proposals that are well aligned with the goals of the solicitation.

[Raleigh Martin] 13:25:32
And our general advice since we do. Get quite a few inquiries is when reaching out. Ideally prepare a one-page summary of what you're thinking about.

[Raleigh Martin] 13:25:41
Including how your project responds to the specific requirements of the CAI G program that'll help to inform a more substantive conversation and advice that we will have with you.

[Raleigh Martin] 13:25:54
Okay, so that's the end of the presentation portion of this. In a moment, I'll stop sharing my slides and we'll turn over to a panel of CAIG program officers who are listed here.

[Raleigh Martin] 13:26:07
To ask questions, please use the Q&A feature, in zoom. You may submit your questions at any time and we'll just kind of go through them one at a time as we see them.

[Raleigh Martin] 13:26:18
You can submit your question anonymously if you choose to do so. If you don't click the send anonymously button, your identity will be revealed.

[Raleigh Martin] 13:26:27
And again, as I've noted, this webinar and the Q&A will be recorded and posted on that CAAG program page probably within a week or so.

[Raleigh Martin] 13:26:35
And if you have any questions beyond this webinar today, please reach out. We have a email alias, Ca ig@nsf.gov.

[Raleigh Martin] 13:26:45
That allows us to kind of feel inquiries and kind of send them to the program officer. Best equipped to respond.

[Raleigh Martin] 13:26:50
And we'll try to get back to you on your questions as soon as we can when we receive them by email.

[Raleigh Martin] 13:26:56
So with that, I'm going to stop sharing.

[Raleigh Martin] 13:27:01
And if. I think everyone is here. So we have several program officers. I'll just briefly introduce them.

[Raleigh Martin] 13:27:10
Jodi Meet is from our division of mathematical sciences. Sylvia Spangler is from our division of intelligent information systems and size directorate.

[Raleigh Martin] 13:27:20
Sean Kenan from our Division of Ocean Sciences. Mark Stiglitz from our Office of Polar programs.

[Raleigh Martin] 13:27:28
Joseph Carlin from our division of ocean sciences. He's a, from Division of Earth Sciences.

[Raleigh Martin] 13:27:33
And Eric to we were from division of atmospheric and geospace sciences. So we have broad coverage here today to address your questions.

[Raleigh Martin] 13:27:41
They come through. So now I'm going to pull up that Q&A. Okay, so the first question is.

[Raleigh Martin] 13:27:49
Whether one of the Ko Pi should have people from computer science department. So I'm gonna answer this live.

[Raleigh Martin] 13:27:57
So in general, we're looking for collaborations. That include folks who would identify as being primarily geoscientists.

[Raleigh Martin] 13:28:07
As well as folks who had identify as being primarily experts in AI approaches. Whether those experts are within a computer science department specifically.

[Raleigh Martin] 13:28:18
Or some other related type of department. That's not a specific thing we're gonna be checking for.

[Raleigh Martin] 13:28:26
The important thing is that we see collaborations. Across these just different disciplinary backgrounds and that these are meaningful collaborations with an equal partnership.

[Raleigh Martin] 13:28:35
Among all the project leads.

[Raleigh Martin] 13:28:37
Anything else that any of the other program officers here would like to add to that point?

[Raleigh Martin] 13:28:47
Sylvia, go ahead.

[Sylvia Spengler] 13:28:50
People that do computer science. Tom and a variety of shapes and colors. And and particularly departments.

[Sylvia Spengler] 13:29:01
Especially in the last decade, departments have be found hiring. Cross-disciplinary people.

[Sylvia Spengler] 13:29:09
So what we really care about is the expertise and the demonstrated experts. And it is possible that you might be in the same giant department.

[Sylvia Spengler] 13:29:22
That's quite possible. Or school. And I think it's important to recognize that. So we don't want to harden fast rule about where, you know, what your credentials have to be, but we want to see.

[Sylvia Spengler] 13:29:37
In in your. Publications, etc. Where your expertise is so that we can judge it appropriately.

[Raleigh Martin] 13:29:54
Any other comments people wanted to add to that?

[Raleigh Martin] 13:29:59
And I also see there is a question. I think this is. You know similar to what you were just addressing Sylvia.

[Raleigh Martin] 13:30:05
Can the interdisciplinary project team be at the same department? Again, it's less about the specific departmental affiliation you have.

[Raleigh Martin] 13:30:15
And more about the expertise you bring to bear. You could have 2 people in 2 different departments, but if they're coming from the exact same disciplinary background, That's gonna not be looked upon.

[Raleigh Martin] 13:30:27
Favorably. On the other hand, you could have 2 collaborators from the same department. Who bring very different expertise.

[Raleigh Martin] 13:30:32
So it's really more about the people than it is about their particular affiliation.

[Raleigh Martin] 13:30:40
Any other thoughts on that?

[Raleigh Martin] 13:30:43
Now go to the next question. Cloudbank told us, we have cloudbank box.

[Raleigh Martin] 13:30:51
That they are shutting down completely. Why is it listed here? Okay, that's an interesting question.

[Raleigh Martin] 13:30:59
That's the first that I've heard about, that. I don't know if others on the webinar are aware of that, but as far as we're concerned, Cloud Bank is still alive and well.

[Sylvia Spengler] 13:31:10
I've heard it. I've heard it. I haven't seen it yet. That is, I haven't heard a statement.

[Sylvia Spengler] 13:31:19
About it. But it was Well subscribed, perhaps. Overly enthusiastically, so.

[Sylvia Spengler] 13:31:29
And the hope is that the other kinds of infrastructure. We'll. Will take up the Slack, but I realize that it's a.

[Sylvia Spengler] 13:31:43
It's a gap.

[Raleigh Martin] 13:31:48
So what I'll just say. Proposers is to at this time you know at the time that the solicitation was approved and published Cloud Bank was the next time program and it is, you know, as far as I know.

[Raleigh Martin] 13:32:04
So I wouldn't let that affect. Your proposal preparation, but if you do come across specific questions with Cloud Bank as you were getting ready.

[Raleigh Martin] 13:32:13
To submit, please do reach out to us and we'll provide you with a timely answer on that.

[Raleigh Martin] 13:32:22
Okay, so next question. Regarding the disciplinary focus does space science ie space weather, Earth's magnetosphere, the radiation belts.

[Raleigh Martin] 13:32:32
Jump out as within the area focus. I'm gonna kick that over to Eric since you're in the atmospheric and geos space science and let you answer that.

[Eric DeWeaver] 13:32:42
Yeah, I mean. I would say that, you know, the scope of CAIG is essentially the scope of the core programs of the geosciences directorate.

[Eric DeWeaver] 13:32:53
Since we have a space where the program that should be legitimate. I would point out though that you know, as with all, you know, as with anything labeled weather, you know, there's an operational side of it in this research side of it.

[Eric DeWeaver] 13:33:08
You know, so we would. Preferentially fund. Activities which, focus on the basic size aspects of space.

[Eric DeWeaver] 13:33:17
There's a lot of use of course in the development of schemes that have some predicted value but not all empirical predictions really, serve to advance the basic science understanding of the phenomenon that's being predicted.

[Raleigh Martin] 13:33:37
Thanks, Eric. Okay, we'll move on to the next question. Does the proposal scope cover land processes interacting with vegetation?

[Raleigh Martin] 13:33:48
I, how climate interacts with vegetation. Well, given the climate emphasis here, maybe that's another question for you, Eric.

[Eric DeWeaver] 13:33:59
Yeah. That's a tricky question.

[Eric DeWeaver] 13:34:04
Being that this is the geosciences directorate. We don't really get involved with the biology physiology.

[Eric DeWeaver] 13:34:14
Ecosystem dynamics of vegetation. That said, you know, we do sort of support research.

[Eric DeWeaver] 13:34:24
On land atmosphere interactions which affect climate You know, so something to do with, evapo transpiration and how that affects atmospheric humidity and other aspects of.

[Eric DeWeaver] 13:34:37
Of land atmosphere coupling, that would be something that we actually do fund. And so that would be a topic that would be covered here.

[Eric DeWeaver] 13:34:45
We can follow up if you like. I mean, I don't know if there's a, if that's a sufficient explanation or not.

[Raleigh Martin] 13:34:52
Mark, please go ahead.

[Marc Stieglitz] 13:34:57
I was gonna say in the Arctic where it's a place rather than a discipline we do fund quite a bit.

[Marc Stieglitz] 13:35:05
Of ecological work and it's usually the ecological work that we co-found with, the biology directorate.

[Marc Stieglitz] 13:35:16
And that does focus on questions of phonology, carbon nitrogen cycling. So.

[Marc Stieglitz] 13:35:22
So and obviously land happens for interactions, plant soil interactions. So that is the one. Caveat in a sense to that.

[Marc Stieglitz] 13:35:34
And as such it's the kind of things that our core programs like Project Natural Sciences would fund.

[Raleigh Martin] 13:35:43
Thank you, Mark. And I just wanted to add one thing here. We get inquiries related to, for example, the use of remote sensing to or application of AI based techniques to analyze remote sensing imagery.

[Raleigh Martin] 13:35:59
To, you know, monitor changes that are occurring at the land surface. The important thing to remember is If that kind of analysis is being done primarily to inform an operational.

[Raleigh Martin] 13:36:13
Are an applied use. That is probably outside of scope of the types of research questions that are typically considered.

[Raleigh Martin] 13:36:20
As of interest the geosciences, instead there really has to be a sense that there is some kind of fundamental science question that is being enabled through the development of these approaches.

[Raleigh Martin] 13:36:33
So just something to keep in mind there.

[Raleigh Martin] 13:36:38
Okay, I'll move on to the next question. Would you please clarify how CAIG is different or complementary to the ge Informatics program.

[Raleigh Martin] 13:36:48
So I'll answer that because I'm a program officer also for ge Informatics. So CAIG is a program in our division of research innovation synergies and education rise.

[Raleigh Martin] 13:36:59
In the geosciences directorate in collaboration with SIZE director at an NPS directorate.

[Raleigh Martin] 13:37:05
So it's really meant to cover The full scope of the geosciences plus. Computer science plus mathematical science.

[Raleigh Martin] 13:37:15
And there's a strong focus on this development of advanced AI approaches in tandem with addressing research bottlenecks in the geosciences.

[Raleigh Martin] 13:37:28
In contrast, the geoformatics program first, it's specific to the division of Earth Sciences.

[Raleigh Martin] 13:37:35
So that's smaller than the whole scope of the the director for geoscience. So excludes.

[Raleigh Martin] 13:37:40
Atmosphere, polar, ocean science. Also ge Informatics is much more specifically about supporting enabling cyber infrastructure or science research and education.

[Raleigh Martin] 13:37:54
And so Geinformatics is, you know, there's plenty of projects we support in that program that have nothing to do with AI and that's completely fine.

[Raleigh Martin] 13:38:01
So I think both in terms of the topics that are being covered and also the disciplinary scope, there's quite a bit of difference.

[Raleigh Martin] 13:38:07
Again, I would encourage you to go to that geoci web page. And there's actually a full listing a lot of a lot of these programs and how.

[Raleigh Martin] 13:38:15
Fit together.

[Raleigh Martin] 13:38:20
Okay, so. Yes. So there's just a question about sending a link to the PowerPoint.

[Raleigh Martin] 13:38:26
We will post this. Probably within a week or so, you know, takes a bit of time to get the website updated.

[Raleigh Martin] 13:38:31
But that is our intent to do that. Let's see. How will the review process work?

[Raleigh Martin] 13:38:38
Will it be peer review or panel or both? So. In general, we anticipate a combination of both individual reviews.

[Raleigh Martin] 13:38:50
By individual experts. As well as panel review composed of a group of experts who convene for a couple of days and look at a batch of proposals and we're going to in general try to get expertise that's matched to the collaborative interdisciplinary nature of this solicitation.

[Raleigh Martin] 13:39:11
So you'll have. Experts who are kind of focused more on the geoscience disciplinary aspects.

[Raleigh Martin] 13:39:17
They'll have experts who are focused more on the AI. Methods aspects of the proposals really try to get a well-rounded, set of expertise to evaluate the proposals to inform.

[Raleigh Martin] 13:39:30
The recommendations that we end up making as program officers for what ends up getting supported. Anything folks will want to add to that.

[Raleigh Martin] 13:39:43
Okay. Next question, is it expected and courage considered important to have AI research scientists on the team?

[Raleigh Martin] 13:39:54
I'll give my initial answer. I would say yes. I think it's not sufficient.

[Raleigh Martin] 13:40:01
If you had a team that's just folks who are primarily identified as geoscientists with strength in AI driven approaches.

[Raleigh Martin] 13:40:10
I don't think that's going to meet the bar of the partnership requirement for the solicitation.

[Raleigh Martin] 13:40:16
We really do want folks who come to this. Both with, you know, one at least one member of the team lead who is a geoscientist.

[Raleigh Martin] 13:40:25
And another team lead who comes from you know expertise in one of those AI driven approaches but you know who exactly that person is and how exactly they Identify, either we expect some variation.

[Raleigh Martin] 13:40:38
So I don't know if others wanted to provide any further clarifying comments there.

[Eric DeWeaver] 13:40:44
Well, I guess I would just say that. You know, just as computer scientists work in a number of different departments, you also find people with expertise and artificial intelligence sort of scattered throughout academia.

[Eric DeWeaver] 13:41:00
So I don't know that we're necessarily going to be. Bound by a specific credential, cause I'm not quite sure what that credential would be.

[Eric DeWeaver] 13:41:09
But certainly some demonstration of you know. Expertise in artificial intelligence and you know ability to kind of you know hold up that that half of the partnership, I think that is critical.

[Sylvia Spengler] 13:41:30
It's also important and critical to be able to identify what's innovative. And. And if there's an innovative use of AI.

[Sylvia Spengler] 13:41:43
As opposed to a development of something. If it's, if it's a use.

[Sylvia Spengler] 13:41:49
Arguments. Then you have to be able to identify. What is specifically innovative?

[Sylvia Spengler] 13:41:58
And that often comes more with the more focused expertise.

[Raleigh Martin] 13:42:08
And I just wanted to reiterate here. We have a merit review process. That acknowledges that no one person is going to be able to evaluate this completely.

[Raleigh Martin] 13:42:19
Yeah, we're gonna get reviewers with a range of expertise and ultimately it's incumbent on the proposer to convince the reviewers through the, you know, limited 15 pages you get in the project description of your proposal.

[Raleigh Martin] 13:42:32
And the other proposal documents that you have a convincing plan. For how the team you've assembled is establishing a meaningful partnership.

[Raleigh Martin] 13:42:41
How are you are achieving that that geoscience advancement, how you are achieving that AI impact.

[Raleigh Martin] 13:42:47
You know, in the end of the day, it's gonna be something of a judgment call here.

[Raleigh Martin] 13:42:52
Among the reviewers so the more clearly you can articulate that in your proposal don't assume that because we're gonna have a breadth of of a of reviewers don't assume higher knowledge of kind of specific aspects of what you're doing.

[Raleigh Martin] 13:43:08
You know, lay it out very explicitly. And convince the reviewers that you are indeed achieving that.

[Raleigh Martin] 13:43:13
Partnership that we're looking for.

[Raleigh Martin] 13:43:19
So a question here. I think this was already answered. Do some of the Pi's have to be in geo-related departments?

[Raleigh Martin] 13:43:26
Can we have all Hey, I related PIs with some of them having prior expertise in GEO.

[Raleigh Martin] 13:43:32
So again, you know, I think There's not like a formal requirement here in terms of departmental affiliation.

[Raleigh Martin] 13:43:39
Or in terms of research. Specific research background. I think it's really a Providing a convincing demonstration that there is this true collaboration where you're bringing together expertise from disparate areas.

[Raleigh Martin] 13:43:53
Both in the geoscience and in AI.

[Raleigh Martin] 13:43:59
Next question is fieldwork expected for this project and what is the expected balance between field work? An AI data processing.

[Raleigh Martin] 13:44:09
I think the short answer to that is no. There is a lot of geoscience related research.

[Eric DeWeaver] 13:44:13
Yeah.

[Raleigh Martin] 13:44:17
It doesn't. Include field work so just as we wouldn't expect. A field work component in any other geoscience program.

[Raleigh Martin] 13:44:24
I don't think we explicitly expected here. I think. You know our general advice is that the work plan should be suited to the project goals.

[Raleigh Martin] 13:44:33
But any other thoughts on that?

[Raleigh Martin] 13:44:40
Okay. So I have 2 questions here. How many years will the project last? And may I know the timeline, the earliest project starting a notification date?

[Raleigh Martin] 13:44:52
So in terms of the project duration, the solicitation states of that can be up to 3 years that you can propose for.

[Raleigh Martin] 13:44:59
Support. In terms of the review timeline, so our deadline is March fifteenth.

[Raleigh Martin] 13:45:08
At NSF, we typically endeavor to have Notification date within 6 to 9 months after the proposal is received.

[Raleigh Martin] 13:45:17
I'm hoping, you know, our goal or standard of service is to be closer to that six-month timeline, but you know, really depends on a variety of factors.

[Raleigh Martin] 13:45:25
Number proposals received availability of funding, etc. So that's just kind of the general guidance we can give.

[Raleigh Martin] 13:45:31
There.

[Raleigh Martin] 13:45:36
To the panel, can you provide some illustrative examples of research questions in ocean and earth sciences?

[Raleigh Martin] 13:45:42
There would be a good fit for CAI G. So do our ocean or earth scientists on the panel want to answer that.

[Raleigh Martin] 13:45:53
I know in general, I think we're hesitant to pick research questions for proposers. You know, it's really out to you to come to us with your good ideas.

[Raleigh Martin] 13:46:01
You know in general I would say look at the program descriptions for our core disciplinary programs and division of our science division of ocean sciences.

[Raleigh Martin] 13:46:10
I think that should give you a good starting point for disciplinary breadth, but. Happy to hear from.

[Raleigh Martin] 13:46:16
Eva, Shawn or Joe, if you wanted to talk about those fields.

[Raleigh Martin] 13:46:25
And you know, one thing that Folks can always do is Go to, the NSF Award search.

[Raleigh Martin] 13:46:33
Search for some key words that are of interest. Look at projects that have been funded. There, you know, there's not a base of projects that have prior been funded by CAIG, but there is, you know, a large base of projects that have been supported through the directed or or director for geosciences and so You know, that's always a resource.

[Raleigh Martin] 13:46:52
To kind of look at themes that are you know tending to come up in Nsfunded research.

[Raleigh Martin] 13:47:03
Okay, next question. Our team contains an AI expert and a geoscientist. Generally speaking, who is preferred to be the leading PI?

[Raleigh Martin] 13:47:14
So in general. You know, PI is actually a formal designation. For NSF that implies that you are kind of an official conduit between your organization.

[Raleigh Martin] 13:47:28
And NSF in matters related to your proposal. If I understand correctly and others, please correct me if I'm wrong.

[Raleigh Martin] 13:47:35
There's no formal distinction. In ansf's eyes between who is listed as main PI and who is listed as KPI.

[Raleigh Martin] 13:47:42
Other than basically the API is the first person whose name appears. On that project. Cover page and abstract.

[Raleigh Martin] 13:47:50
But aside from that, you know. I think our general advice is to pick. The teams and the roles that make the most sense for the project.

[Raleigh Martin] 13:47:59
Any other thoughts on that phone? Folks here.

[Sylvia Spengler] 13:48:06
It's an effort to gain the system and it's that's too easy. What you have to grapple with the problem.

[Sylvia Spengler] 13:48:14
Of bringing together your AI elements and your Geo elements. And it's not there isn't a nice rule for it.

[Sylvia Spengler] 13:48:26
Sorry. I know AI is a lot about rules, but this isn't one of them.

[Raleigh Martin] 13:48:33
Very good. I mean, one thing I will say, you know, we have this. Expectation that there's 2 to 3 leads senior personnel which conclude Pi's and I think one thing that could affect your proposal is if all of those folks who are identified as project leads are coming from a similar research background.

[Raleigh Martin] 13:48:55
That is going to potentially adversely affect the review of your proposal. But in terms of, you know, specifically like who among those 2 to 3 is lead PI versus CoPI.

[Raleigh Martin] 13:49:06
That's really, you know, up to you, a decision to make.

[Raleigh Martin] 13:49:13
Next question, does the solicitation emphasize more on the innovative AI method? If we plan to combine our domain knowledge with AI to improve some existing application.

[Raleigh Martin] 13:49:22
Does this fall into the scope of this solicitation call? So this is, I'd like to hear some perspectives from, my colleagues here.

[Raleigh Martin] 13:49:33
I think what we're trying to avoid in the solicitation is simply taking an AI based method.

[Raleigh Martin] 13:49:40
An analysis machine learning analysis approach that already exists in applying it. To a geoscience problem.

[Raleigh Martin] 13:49:49
There has to be something beyond just taking the method off the shelf. Maybe the method is 80% of the way there.

[Raleigh Martin] 13:49:54
And there's a 20% value add that you need to provide to advance that method or maybe it's something brand new, you know, a physics informed.

[Raleigh Martin] 13:50:01
Ai method. Again, you know, it's going to be a matter of you convincing the reviewers where that innovation in AI impact lies.

[Raleigh Martin] 13:50:12
So it's hard to give a definitive answer. And again, you know, it depends on you providing that convincing demonstration.

[Raleigh Martin] 13:50:16
In your proposal. But I think that's like roughly what we're looking for here. But I would welcome responses from my colleagues.

[Jodi Mead] 13:50:25
So I could speak up a bit here. It may be the case that an existing application. That when using AI methods you discover that.

[Jodi Mead] 13:50:37
There are issues that need to be addressed that others haven't thought of because now you're using the AI and the context of a geoscience application.

[Raleigh Martin] 13:50:50
Thanks, that's a really good point, Jodi.

[Raleigh Martin] 13:50:56
Any other thoughts on this?

[Sean Kennan] 13:50:59
Yeah, I'll just add a little Raleigh. I mean. There have been several questions that kind of touch on the same issue, right?

[Sean Kennan] 13:51:06
So. I would just reiterate that, the purpose of this call is really focused on the collaborations, right?

[Sean Kennan] 13:51:13
Bringing together. People from different fields. So even though as has been repeated several times. We don't have a litmus test.

[Sean Kennan] 13:51:21
For looking at people's titles or departments. It should be evident when you put your project together that you're bringing people together that wouldn't have otherwise necessarily.

[Sean Kennan] 13:51:31
Have been working together and in particular that the The nexus of the science problem. Is not something that would have generally fit wholly within an existing core program and stuff, right?

[Sean Kennan] 13:51:44
So if you're. You know, if you're geoscience problem could be submitted.

[Sean Kennan] 13:51:48
You know, to one of the existing divisions under the open solicitations, then it's probably not appropriate for this call.

[Sean Kennan] 13:51:56
This these are this is really meant for projects that wanna take an innovative approach. And again, the emphasis can be a little bit more on the AI or a little bit more on the geoscience.

[Sean Kennan] 13:52:08
But it should really be that need to bring people together in a non-traditional way to look at problems in a new light.

[Sean Kennan] 13:52:17
Or innovative from at least one of the sides of the aisle. So, you know, all of these questions, it's a little bit hard for us to answer because it'll depend on the proposals that come in but hopefully from all these different answers, you can all get a picture of, kind of what the intent of the solicitation is.

[Sean Kennan] 13:52:35
And again, just to reiterate that you can reach out to any of us separately. For your particular project if you're still confused or have questions on the viability of your.

[Sean Kennan] 13:52:45
Idea.

[Raleigh Martin] 13:52:49
Thanks, John. That's very helpful. Any other comments on this?

[Raleigh Martin] 13:52:58
Yeah. I'll just second. Please reach out to us. Send us a one page summary and we'll, provide.

[Raleigh Martin] 13:53:06
The best advice we can.

[Raleigh Martin] 13:53:11
Next question, geosciences includes various domains. Do you encourage the proposal to cover wide ranges of domain or specific disciplines?

[Raleigh Martin] 13:53:20
So. In general, you know, I think we are looking for the disciplinary scope that you know, is appropriate for a well-formed project.

[Raleigh Martin] 13:53:31
There is definitely not the expectation that you're covering the full range of geosciences. You cover something you know very specifically within Gophysics or biological oceanography or you know space weather or it could be, you know, across a broader range of topics.

[Raleigh Martin] 13:53:50
I think the most important thing Is that the topics are of interest in some way to the geosciences?

[Raleigh Martin] 13:53:56
That there is some partnership that leads to innovation and AI driven approaches to those questions. Hands that.

[Raleigh Martin] 13:54:06
Overall the project holds together and you have a convincing plan for how you're gonna achieve those goals.

[Raleigh Martin] 13:54:11
But I'll let others answer if they have further thoughts.

[Eric DeWeaver] 13:54:17
Yeah, I mean.

[Eric DeWeaver] 13:54:20
Hi, I might hope that, you know, there are people out there who have. AI, you know, tools and methods that would be broadly applicable across, the geosciences.

[Eric DeWeaver] 13:54:33
And I think in those cases, No, maybe it's worth considering that. For the amount of money that we can provide through the solicitation, you're not going to be able to cover all possible applications of the methodology or whatever it is you've got.

[Eric DeWeaver] 13:54:46
And so maybe the interesting question then becomes, what's the best way to prove the concept?

[Eric DeWeaver] 13:54:52
You know, if this is broadly applicable, then they must be more than one test case that would suffice to demonstrate, you know, the value of this tool for applications.

[Eric DeWeaver] 13:55:02
No, and put your best foot forward in terms of, you know, finding. The right application, you know, the good one that, you know, would presumably.

[Eric DeWeaver] 13:55:10
Review the best solicitation.

[Eva Zanzerkia] 13:55:16
This kind of gets to the next question. Just remember that your proposal is going to be, this is a competition across the geosciences.

[Eva Zanzerkia] 13:55:26
So it's incumbent on. Proposers to justify why they're scientific problem, even if it's domain specific.

[Eva Zanzerkia] 13:55:35
Is an important one to address. In a way, that a broad general audience of AI researchers and geoscientists would understand.

[Raleigh Martin] 13:55:45
And I think just to read the question that Eva's responding to, the question was, given the wide scope of Earth science, should proposals be written in the style of a proposal, should proposals be written in the style of a proposal to disciplinary program within or said more accessible for a wider audience.

[Raleigh Martin] 13:56:00
So. Yeah, repeating,'s point basically. You know, because of the breadth of topics we expect to see.

[Raleigh Martin] 13:56:08
In this competition. Expect a similar breadth of expertise. So, you know, if you have a tectonics focused.

[Raleigh Martin] 13:56:17
Proposal. Don't expect you're gonna have a majority of tectonic. Experts on the panel.

[Raleigh Martin] 13:56:23
You're gonna have folks, you know, these are well-trained, educated scientists. From a range of backgrounds, but they may not be familiar with your specific research question.

[Raleigh Martin] 13:56:32
And so You know, it's incumbent on you to really lay that out for the reviewer.

[Raleigh Martin] 13:56:36
To make an assessment of your proposal.

[Raleigh Martin] 13:56:44
Does next question does NSF have a mechanism to assist state non federal geological surveys?

[Raleigh Martin] 13:56:51
And joining CAIG research teams as unpaid collaborators.

[Raleigh Martin] 13:56:57
Not quite sure how to. Answer this. You know, I would say In general, you know, state geological surveys, you know, if they have interests in the topics that are being proposed even if they're not requesting funding.

[Raleigh Martin] 13:57:15
From an SF. You know, certainly encourage you to think about ways to describe their role in the project, in the project description and then demonstrate that.

[Raleigh Martin] 13:57:24
Intended unfunded collaboration through including letters of collaboration. I think that's basically the approach that one would take, but others wanted to clarify that.

[Eva Zanzerkia] 13:57:37
Yeah, I think if the question is. You know, there's a, an, a set of eligible types of institutions, even though geological surveys are not included in that list, it does not mean that they can't be collaborators in an unpaid way.

[Eva Zanzerkia] 13:57:54
And as Raleigh noted, you should submit a letter of collaboration and clearly articulate what your role on the project would be in the project description.

[Raleigh Martin] 13:58:04
Go ahead, Sylvia. I think you're muted.

[Sylvia Spengler] 13:58:07
I am muted and also to use the facilities part. Of the solicitation. To further describe what resources.

[Sylvia Spengler] 13:58:19
Such a Such a state group would bring. To the project. Don't forget to be able to use that.

[Sylvia Spengler] 13:58:29
How it's used can be described in the solicitation, but actually what it is because you will have a broad audience is best described in the facility statement.

[Raleigh Martin] 13:58:42
Yeah, this is I would encourage folks with unfunded collaborators to look at that section of the proposal and award policies and procedures guide the Patchy because I think it lays it out very clearly.

[Raleigh Martin] 13:58:53
There are kind of 3 legs to the stool in terms of how you describe an unfunded collaboration.

[Raleigh Martin] 13:58:58
The first, you know, your page limited here, but in the project description describing the role, the overall role of that unfunded collaborator in the project.

[Raleigh Martin] 13:59:08
The facilities equipment and other resources document. Describing the resources that that unfunded collaborator is bringing to the table.

[Raleigh Martin] 13:59:16
And then the letter of collaboration, which is a boilerplate statement. That is specified in the is the kind of affirmation of the intent of that entity.

[Raleigh Martin] 13:59:26
To be an unfunded collaborator as described in those first 2 parts that I just mentioned. So look at that language.

[Raleigh Martin] 13:59:32
I think Eric, you put it in the chat, but I think you might have sent it only to the to us.

[Eric DeWeaver] 13:59:39
Oh yeah, so don't know how to do that.

[Raleigh Martin] 13:59:39
Sorry, I'm gonna. I'm gonna send it to everyone. I just put it in the chat right there.

[Eric DeWeaver] 13:59:43
That's just the template for, you know, how to actually write the letter of collaboration.

[Eric DeWeaver] 13:59:49
There's other references to collaboration letters in the Yeah, I mean, I think the key question there is.

[Raleigh Martin] 13:59:54
So next question does enabling onboard space AI for climate applications within the scope. I think this is one of those questions where probably we would need to talk to you directly to kind of get a better sense of what it is specifically you're looking to do.

[Raleigh Martin] 14:00:13
Unless others wanted to try to address that question.

[Raleigh Martin] 14:00:19
Yeah, I think reach out to us and we're happy to discuss your specific idea with you.

[Raleigh Martin] 14:00:28
If the proposal is directly related to operation. Does that tone down the scientific merit? I will say, you know, as kind of a general statement.

[Raleigh Martin] 14:00:39
If your goal is to innovate in AI primarily to inform an operational means say like weather prediction is Eric.

[Raleigh Martin] 14:00:51
Pointed out earlier that is probably not going to be the best fit to the program. That being said, you know, in general.

[Raleigh Martin] 14:00:59
Please reach out to us. We can kind of, discuss with you your specific topic. But also you know look at those both the the standard review criteria broader impacts and intellectual merit and also the solicitation specific review criteria.

[Raleigh Martin] 14:01:13
And try to look at it from the perspective of a reviewer. If you were a reviewer looking at what it is you're proposing.

[Raleigh Martin] 14:01:21
Would you see those, those specific requirements of the solicitation being addressed? I think that's kind of you know the general mindset to take when thinking about how to put your proposal together but welcome answers from others.

[Raleigh Martin] 14:01:40
Okay, hearing none. The next question is. Is applied AI based groundwater research allowed for submission?

[Raleigh Martin] 14:01:49
I think similar answer here. You reach out to us, we can talk to you about your idea. Hey, I think in general, if the goal were to just quantify the availability of groundwater resources without kind of scientific question behind it that's probably not gonna do so well.

[Raleigh Martin] 14:02:06
But it's gonna kind of get to the details of the research topic. So. You know, encourage you to reach out to discuss with us.

[Raleigh Martin] 14:02:16
Okay, next question. How innovative AI technology needs to be for geoscience inspired problems? Does it require it to be novel to the geoscience field or innovative in AI community?

[Raleigh Martin] 14:02:30
So, you know, I know we've gotten multiple of these questions about the innovation in AI.

[Raleigh Martin] 14:02:36
This is a requirement of the solicitation. Again, you know, something that. Is gonna be a bit of a judgment call and we're gonna depend heavily on the expertise of our reviewers to assess.

[Raleigh Martin] 14:02:48
But I think in general it does need to go beyond just novelty to geoscientists.

[Raleigh Martin] 14:02:54
I think there has to be some way in which a person who's an expert in AI could say.

[Raleigh Martin] 14:02:59
Here is something that Yeah, we haven't done before and through this project we're gonna do something new.

[Raleigh Martin] 14:03:05
Maybe it's you know, not, you know, a completely new method. Maybe it's just a new extension of an existing method, but there hat does need to be something new about the.

[Raleigh Martin] 14:03:13
The approach, I think to be successful, but, Sylvia, I see your hand raised.

[Sylvia Spengler] 14:03:18
So I put up my hand because sometimes what happens is that the data in a specific field have quirks.

[Sylvia Spengler] 14:03:30
To be kind. That it's incomplete in a given way or that it's timing is highly variable anyway.

[Sylvia Spengler] 14:03:42
There's something about the data that's different. And it's different for this specific domain because of the way the domain is.

[Sylvia Spengler] 14:03:49
And and enabling AI that you know works in something to be able to be used. In that circumstance is an innovation, but you have to be clear about.

[Sylvia Spengler] 14:04:02
You have to be clear about what's driving the innovation in AI and then why it's important for the geoscience.

[Raleigh Martin] 14:04:18
Sorry, I was on mute. Thank you, Sylvia. Next question is developing methods or surrogate to speed up a process in geoscience using AI within the scope.

[Raleigh Martin] 14:04:28
Think again this is gonna be kind of devil in the details. Please reach out to us. Please convince the reviewers.

[Raleigh Martin] 14:04:36
I think in general there are definitely examples of for example AI for surrogate models that are of interest to geoscientists and are pushing the boundaries of AI.

[Raleigh Martin] 14:04:48
But, you know, it's gonna come down to your specific proposal idea. And how well you can convince the reviewers of what you're doing being innovative and and meaningfully addressing geoscience needs.

[Eric DeWeaver] 14:05:06
What new science will you enable by doing that? You know, certainly if we could snap our fingers and make, you know, 1,000 year long simulations that, you know, one kilometer resolution and blah, blah, blah, then you would be able to do a lot of things that we kind of that would be good things to do.

[Eric DeWeaver] 14:05:27
And so figuring out, you know, what is that set of things that will now become possible, that we're not previously possible.

[Eric DeWeaver] 14:05:33
Because of the constraints, I think that would have to be a big kind of, you know, selling point of the proposal.

[Raleigh Martin] 14:05:43
Thanks, Eric. Next question. We currently have a team of a domain geoscientist, an applied mathematician with some ML applications.

[Raleigh Martin] 14:05:54
The Goscientist who has published extensively on ML applications. Would you say this team is appropriate or should we include someone who is more strictly on the basic AI side?

[Raleigh Martin] 14:06:05
So this is again one of those kind of devil in the details questions. In a way, it's going to be up subject to the interpretation of our reviewers.

[Raleigh Martin] 14:06:17
In the context of the project, whether this Contails a meaningful partnership across fields. But, that said.

[Raleigh Martin] 14:06:26
I don't know if any folks here wanted to say anything further about this.

[Eric DeWeaver] 14:06:32
Well, I mean, I, you know. In my experience, which is limited. Most of the AI effort is in fact on machine learning.

[Eric DeWeaver] 14:06:41
And, Bye that I would need to say that, you know. If you have a team member who has expertise on machine learning, you know, and some way of demorating that expertise and publications certainly count.

[Eric DeWeaver] 14:06:57
That I would think that that's going a long way towards, you know, having a meaningful collaborator between someone doing geos, people doing geoscience and people doing.

[Eric DeWeaver] 14:07:07
You know, AI is sort of very large. But I think, yeah, there could easily be devils.

[Eric DeWeaver] 14:07:12
In the details there as well.

[Raleigh Martin] 14:07:16
You know, I think part of this gets to the meaningfulness of the collaboration. You know, we don't want this, we don't want projects where, you know, it's fundamentally a geoscience research question.

[Raleigh Martin] 14:07:28
You're kind of renting an AI expert to do one specific task. We want these to be research questions where both The geoscientists involved.

[Raleigh Martin] 14:07:36
And that the AI experts involved, whether they're mathematicians, whether they're computer scientists, are excited by the research problems.

[Raleigh Martin] 14:07:43
That are being put forward. If they're collaborating simply to, you know. Consult on some specific method but they don't see this as meaningful and their research.

[Raleigh Martin] 14:07:54
I think that's a red flag. I think. That all the collaborators involved should see something within their own.

[Raleigh Martin] 14:08:00
Field where they see some exciting innovation. And only by coming together with folks from different fields. Are they going to meaningfully, you know, achieve, research objectives in their respective fields?

[Raleigh Martin] 14:08:14
So I think that's kind of, you know, a bigger picture way to think about this.

[Raleigh Martin] 14:08:20
Next question, will there be any interest in receiving explainable or counterfactual AI? I-based application proposals?

[Raleigh Martin] 14:08:30
I think this is again comes to Convince the reviewers. Why is using explainable counterfactual AI you know, where is the innovation and how is this going to?

[Raleigh Martin] 14:08:42
Over come geoscience polymics.

[Eric DeWeaver] 14:08:47
Let me a certain sense that I think explainable AI makes sense because you know the common complaint that we have about AI is that nobody really knows how these things work.

[Eric DeWeaver] 14:08:59
You know, and if you have a black box that put that spits out answers given inputs and no one knows why.

[Eric DeWeaver] 14:09:03
How can such a thing meaningfully contribute to scientific understanding. No, and so to the extent that the explainability component of this.

[Eric DeWeaver] 14:09:13
Makes the technique more relevant to science and more sort of scientifically credible, actionable, so on and so forth.

[Eric DeWeaver] 14:09:22
Then I would think that's a reasonable thing to do. But again, you know, I think, you know, just by saying that it's explainable AI, you know, there could be devils in the details there as well.

[Sylvia Spengler] 14:09:35
Still have to have the innovation. Either that or the clear and the clear integration with what with the genoscience question is more model or fill in the blank floor.

[Sylvia Spengler] 14:09:48
Or decision process for things you should do. Climates or whatever. We still need the details.

[Raleigh Martin] 14:10:03
Thanks, Eric and Sylvia. Next question, what is the guideline on maximum total budget in each proposal over the project?

[Raleigh Martin] 14:10:11
Is 1 million dollars possible supposing The sufficient research tasks are proposed. So we we intentionally did not include an upper budget limit because We wanted people to basically propose the amount of funds.

[Raleigh Martin] 14:10:28
Requisite to the scope of work that they intended to do. So. You're given the the announced availability of funds you know expected 6 to 10 million dollars.

[Raleigh Martin] 14:10:40
The expected number of projects to be supported 10 to 15. Hey, I definitely think that the project's pushing.

[Raleigh Martin] 14:10:47
Up to or over the 1 million dollar mark are going to, you know, cause some strain in our available budgetary resources.

[Raleigh Martin] 14:10:54
But again, that's not like a hard and fast limit. You know, the most important thing is to propose the budget.

[Raleigh Martin] 14:11:01
To the scope of work that's being proposed noting. Those were requirements for, you know, up to 3 years.

[Raleigh Martin] 14:11:08
Having you know a team of 2 to 3 leads that bring in collaborative expertise. But you know, the people cost different amounts of money depending on what they do, where they're based.

[Raleigh Martin] 14:11:19
What their role is so you know that's why we're not trying to dictate a specific bunch of the people proposed to.

[Raleigh Martin] 14:11:29
Any other thoughts on budgets?

[Raleigh Martin] 14:11:37
And I'll just say with budgets, you know, that the other thing we encourage.

[Raleigh Martin] 14:11:41
You know, think creatively about using some of those shared computing resources like the access program. Keep an eye out.

[Raleigh Martin] 14:11:49
You know, hopefully that then they're reached pilot resource will be coming out. The announcement there soon in the coming days.

[Raleigh Martin] 14:11:54
So, you know, to the extent possible, especially on the computing side, tapping into those existing resources.

[Raleigh Martin] 14:12:01
Will really be important. Of course, as we know, you know, the most expensive part of doing research is the people.

[Raleigh Martin] 14:12:07
And we can't get around the fact that people's time costs money. But you know, we're possible for some of those other resources.

[Raleigh Martin] 14:12:14
Look into what's out there that you can tap into.

[Raleigh Martin] 14:12:19
Next question of all API can be on 2 separate proposals. Does that affect the chance of both proposals being funded?

[Raleigh Martin] 14:12:29
So I would say that the short answer to this is no. That we really are looking for.

[Raleigh Martin] 14:12:36
Those proposals that are at the leading edge in terms of addressing the goals of the solicitation and doing so in a manner.

[Raleigh Martin] 14:12:46
That's, you know, very clearly thought out and has a high potential for success. That being said, you know, we do consider a portfolio balance as one of many considerations when making our award decisions.

[Raleigh Martin] 14:13:01
So I couldn't plausibly say that, you know, PI affiliation with multiple proposals is immaterial.

[Raleigh Martin] 14:13:08
But I can say reasonably that that's not going to be the primary consideration of the funding decisions. Doing a any folks wanna add anything to that.

[Raleigh Martin] 14:13:21
Okay, next question. Can team of 3 Pi's include 2 faculty from one university and third faculty from a second university.

[Raleigh Martin] 14:13:30
HPI is from different. Department spanning computer science or science and civil engineering. So, you know, in general, as we've said, The university and departmental affiliations I think is less material than the, kind of disciplinary backgrounds that, the partners leads bring to the table.

[Raleigh Martin] 14:13:59
Next question, is the team required to make contact with the program directors with one page summary before their proposal submission?

[Raleigh Martin] 14:14:07
The only proposers who must get in touch with us in advance of submission are folks from FFR DCs.

[Raleigh Martin] 14:14:15
Otherwise, though it's not required for you to get in touch with us, we would very strongly do encourage you to do so.

[Raleigh Martin] 14:14:22
This way we can kind of make sure you're on the right track. In terms of your project ideas.

[Raleigh Martin] 14:14:28
We, you know, we don't want to see people preparing proposals that are completely off base and not aligned with the goals of the solicitation.

[Raleigh Martin] 14:14:35
So we do really encourage you. To get in touch, but again, that's not a requirement except for those proposing for FFR DC's.

[Raleigh Martin] 14:14:46
Next question, will it significantly help the proposal to also propose teaching Zoom? And, of course, with the new innovation to neighboring partner universities.

[Raleigh Martin] 14:14:56
So, you know, as described in the solicitation and in my slides, The capacity building component of enabling more people to use AI, whether it's technical capacity or human capacity.

[Raleigh Martin] 14:15:09
You know, that the training and workforce component. He has a goal of the proposal. I think the workforce development alone is not enough to carry a proposal submission.

[Raleigh Martin] 14:15:22
We do also need to see that innovation component. But that is certainly an element that we do. You know, that is welcome, within proposals.

[Raleigh Martin] 14:15:30
And so that is definitely something that I think You know, the specifics of teaching a zoom ML course, you know, it's gonna depend on What do you expect to accomplish?

[Raleigh Martin] 14:15:39
With that? How well thought out is the plan to do that. So, you know, we can't say a priori whether that works, but you know in general that type of activity.

[Raleigh Martin] 14:15:48
Is compatible with the solicitation.

[Raleigh Martin] 14:15:52
And but I welcome because we haven't really had many questions on education. So I don't know if others had thoughts about this aspect of the solicitation.

[Raleigh Martin] 14:16:08
Okay, hearing none. Yeah, let's see. Thank you for doing the solicitation.

[Raleigh Martin] 14:16:16
To do were you gonna say something Sylvia?

[Sylvia Spengler] 14:16:19
I did. I wanted to say that we specifically say when we talk about solicitations specific review criteria.

[Sylvia Spengler] 14:16:28
In the part about does it benefit all the participants? We specifically call out. Cross-training of students and other researchers.

[Sylvia Spengler] 14:16:38
In the methods to be pursued. So it seems to me that, you know, a zoom course is really a handy way to do that.

[Sylvia Spengler] 14:16:49
But. It's not the only way to do it. But answering that part of a specific review criteria is

[Raleigh Martin] 14:17:01
That's a really good point, Sylvia. You know, we recognize that.

[Raleigh Martin] 14:17:07
Especially in AI, the techniques are advancing and changing so quickly. There is going to be the need for, information exchange and cross-training and probably that's gonna likely occur not just at the the student level but it could even be necessary.

[Raleigh Martin] 14:17:24
Among the more senior participants in the project. So, you know, we do wanna see that dissemination of, knowledge.

[Sylvia Spengler] 14:17:30
YouTube courses are are incredibly popular.

[Raleigh Martin] 14:17:40
Excellent. So we just have a few more questions here that haven't been answered yet. I plan to develop a novel AI model which is used for high-resolution Earth system model evaluation.

[Raleigh Martin] 14:17:50
Will that align with the solicitation topic? So I'm gonna throw that at you, Eric.

[Eric DeWeaver] 14:17:58
Well again, I mean I think the The issue of developing, you know, essentially cyber infrastructure. Is going to kind of hinge on, you know, what new science will this enable.

[Eric DeWeaver] 14:18:12
There's a tremendous industry around the world in model evaluation. And. That by itself, I think might.

[Eric DeWeaver] 14:18:23
Have a lower priority. Then, you know, proposals where you're really saying specifically, you know, what are you gonna learn from this?

[Eric DeWeaver] 14:18:34
You know what problems is being applied to. You know, what will, what we know on account of this that we wouldn't have figured out without it.

[Eric DeWeaver] 14:18:43
You know, I think it's certainly true that a tremendous amount of model evaluation is kind of necessary in the background.

[Eric DeWeaver] 14:18:50
You know, so that we can sort of understand how good models are. But I think how good our models is in perhaps the most compelling scientific question that you can ask.

[Raleigh Martin] 14:19:05
Next question, does the proposal have to include post dock or student? Short answer is no.

[Raleigh Martin] 14:19:11
But again, you know, we're, looking at all of the, intellectual merit and broader impacts.

[Raleigh Martin] 14:19:18
And these 3 solicitation specific criteria. So. You know, we're open to all ways of achieving the goals of the solicitation.

[Raleigh Martin] 14:19:26
Whether that's through a postal or student or some other types of project participants is really up to the proposer.

[Raleigh Martin] 14:19:38
Next question. What is the weight given to innovation and teaching and curriculum development? During the review process.

[Raleigh Martin] 14:19:46
So I don't think that that is among the solicitation specific requirements specifically. Innovation and teaching and curriculum development.

[Raleigh Martin] 14:19:56
So. You know, I think, you know, insofar as teaching and curriculum development is addressing one of those.

[Raleigh Martin] 14:20:03
Either intellectual merit broader impacts or solicitation specific requirements. That can be useful but you know, in terms of the balance of, of considerations.

[Raleigh Martin] 14:20:12
For this program. I don't think it's necessary to be. Pursuing innovation and teaching or curriculum development.

[Raleigh Martin] 14:20:19
But Others correct me if I'm mistaken.

[Eva Zanzerkia] 14:20:22
Depending on the extent of the. Curriculum development you're doing, you may wanna consider the cyber-training solicitation that may be a better fit.

[Eva Zanzerkia] 14:20:33
I mean, and we can put a link to that in the chat.

[Raleigh Martin] 14:20:37
Yeah, that's a good point. Yeah, I think as stated in the solicitation that there's this 3 priorities I outlined in the beginning, the first priority that's focused on the AI innovation that there's this 3 priorities I outlined in the beginning, the first priority that's focused on the AI innovation, the second priority focused on the technical capacity and the third priority focused on the human capacity and education.

[Raleigh Martin] 14:20:55
You know, education, advancement of the workforce to use AI based methods in the geosciences is a priority.

[Raleigh Martin] 14:21:02
But that alone is probably not enough to to carry a a proposal in this solicitation. In this competition.

[Raleigh Martin] 14:21:14
Will the question, will the use of data from current NSF projects encouraged? So.

[Raleigh Martin] 14:21:21
I think we're kind of agnostic to, you know, we understand like many AI driven approaches are.

[Raleigh Martin] 14:21:28
Partacing large data sets to train those methods. So certainly to the extent that you can tap into existing data sets to address those needs, that's a good thing.

[Raleigh Martin] 14:21:39
But whether that's coming from current NSF projects, whether it's coming from other data sources I think overall.

[Raleigh Martin] 14:21:47
We would be agnostic, you know, the important thing is that whatever data are being used or fit for purpose and that there's a clear project plan for how to use those data.

[Eric DeWeaver] 14:22:01
Maybe another way to answer the questions. We would to some extent be leaving that to the review process.

[Eric DeWeaver] 14:22:08
Yeah, if we get reviews that say, yeah, this is a great way to use this data. You know that would carry some weight because you would be getting a favorable review.

[Eric DeWeaver] 14:22:17
But it wouldn't be a judgment that's imposed upon this externally by the, program managers.

[Raleigh Martin] 14:22:25
Yeah, that's a really good point.

[Raleigh Martin] 14:22:30
Okay, well, Just in time. We've answered all of the open questions.

[Raleigh Martin] 14:22:38
This webinar is scheduled to end at 2 30. So. At this point I would offer my colleagues if you had any final remarks you wanted to make.

[Raleigh Martin] 14:22:47
Before we close. If there are any final questions from the the webinar participants, please ask them now.

[Raleigh Martin] 14:22:55
Otherwise, we'll call it a day.

[Eric DeWeaver] 14:22:58
I guess I do see one last question as to whether the recording be shared. Not sure what we're doing on that.

[Raleigh Martin] 14:23:04
Yeah, so the answer is yes. We will post the. Slides and the recording on the CAIG website.

[Raleigh Martin] 14:23:15
That the timeline for that, it just, you know, the administratively, it probably take us a few days to get that together.

[Raleigh Martin] 14:23:21
But that is our plan to do that.

[Sylvia Spengler] 14:23:22
And Eva posted the link. To the program page. As the answer to the very first question. So all you have to do if you're in the meeting is scroll back to the answered questions.

[Raleigh Martin] 14:23:39
Or hopefully you're at the program page anyway if you're preparing your proposal because that is also where you find the solicitation.

[Raleigh Martin] 14:23:46
And all of the guidelines for how to prepare your proposal. So we really hope you're actually reading through that in detail.

[Raleigh Martin] 14:23:53
When putting your proposal together. That I can tell you. That all that language in there was written with a great level of intention.

[Raleigh Martin] 14:24:02
We really tried to be concise and not be overly verbose with what we were saying there.

[Raleigh Martin] 14:24:07
So. Each sentence in that solicitation. Is there for a reason? And so please read that carefully.

[Raleigh Martin] 14:24:19
Okay, and I see Eva posted in the chat that the cyber training solicitation that's a separate program.

[Raleigh Martin] 14:24:26
If you're doing a proposal that's really kind of primarily training and workforce oriented for cyber skills.

[Raleigh Martin] 14:24:33
That's a great place to look. I think that deadline is coming up very soon, like early February.

[Raleigh Martin] 14:24:38
So. Maybe too late for a brand new project, but that's an annual competition. To keep an eye on.

[Raleigh Martin] 14:24:50
Any other final comments or? Questions?

[Raleigh Martin] 14:25:00
Okay, well, hearing none, I think we will then formally adjourn. This webinar.

[Raleigh Martin] 14:25:09
Thanks to everyone. For attending today for all the good questions, for your interest in this program and for all of your hard work.

[Raleigh Martin] 14:25:18
In, putting together your proposals and submitting those ideas. We really look forward to seeing some of the great, innovative ideas that you all put forward.

[Raleigh Martin] 14:25:30
And, Yeah. Yeah, thanks again. And also thanks to my colleagues who are here today.

[Raleigh Martin] 14:25:38
I've been providing very helpful responses. And then thanks to our staff support a gabriel bishop who helped us to to host this webinar fall asleep as always so with that, I think we are.

[Raleigh Martin] 14:25:53
Now formally, Sylvia, one more thing. Yes.

[Sylvia Spengler] 14:25:56
Thank you for the for the clapping hands and the thumbs up that bubble up like a spring.

[Sylvia Spengler] 14:26:06
It really makes you smile at the end of a summer.

[Raleigh Martin] 14:26:08
Yeah. Yeah, it makes me feel like I'm on like an Instagram live feed or something on zoom.

[Raleigh Martin] 14:26:18
Okay, great. Well, I think that is done. So I'm going to stop recording.

[Raleigh Martin] 14:26:27
And I think. That's all. I think that the webinar ends when I click the Leave button