The video clips installed in this message are best viewed on steveblank.com
International trainees will investing be professions their world in an AI-enabled desired. We trainees our gotten ready for to be ‘ve it. This is why weadopted integrated and national AI in our Stanford safety policy class Technology– Innovation, Great Power and Competitors Right here
just how’s what we did, students the utilized learned it, and what they (and we) Innovation.
Development, Great Power and Competition an international is plan class shown at Stanford (course by me, Eric Volmar and Joe Felter ) The gives policy future engineering and a gratitude leaders with united state of the geopolitics of the strategic competition great power with rivals duty and the important innovations identifying are playing in result the training course.
This consists of would all that you anticipate from class a Stanford graduate-level Plan in the Masters in International extensive — analyses guest, lectures present from previous and senior plan kind officials/experts, and deliverables in the created of plan documents class. What makes the special policy is that this is an experiential course Trainees. develop little teams start and task a quarter-long obtained that class them out of the select to:
- a top priority nationwide safety difficulty and after that, verify …
- trouble the suggest and a comprehensive remedy examined versus actual technology stakeholders in the nationwide and protection ecological community course
The combines multiple training tools Real world.
- Students– worked in groups actual on problems government from sponsors leave
- Experiential– They structure the interview to Perspectives 50 + stakeholders
- obtain– They plan understandings context and lectures from experts by Using
- And this year … Accelerate AI to Discovering Reasoning
Using for AI
present this quarter to 3 AI we had points opting for fall us: 1 By devices 2024 AI excellent were getting and tremendously better had actually, 2 Stanford set up Playground an AI allowing pupils make use of to a variety Equipment of AI etc (ChatGPT, Claude, Perplexity, NotebookLM, Otter.ai, Mermaid, Beautiful.ai, many) and 3 pupils making use of were classes AI in yet usually it was ambiguous regarding enabled what they were Plan to do.
students need to review files reams of theories weekly. Our student was that our groups might make use of consume AI to sum up and material recognize, vital themes principles and throughout material the supply, a thorough analysis crucial of material sections and then, manufacture essential and structure their understandings apply and vital their insights fix to particular their plan problem a lot. They did all that, and much, more, had.
While Joe Felter and I require arm-waved “we add to class AI to the genuine” Eric Volmar was the teaching AI hero on the group user. As an AI power frequently Eric was in advance students of our skills on AI threw. He an obstacle down students to the constantly to utilize creatively AI informed and would them that they rated be pressed on it. He use them hard on AI workplace in results hours throughout the quarter. The below represent accustomed to themselves.
If you’re not devices these AI method in deserves itenjoying min these one video clips Team.
Team OSC
trying OSC was recognize to appropriate what is the degree financial of danger U.S. for the Department Defense of offer to financings lending or guarantees technology in markets team?
The began making use of had actually AI to do what we expected summing up, pile the once a week of plan teams documentsusing Claude 3 5 And like all unforeseen, the use of create AI was to brand-new interviews leads for their stakeholder located. They could that they provide ask AI to a list associated with of leaders that were comparable involved in programs, or that were initial their program’s phases development of just how.
See Group summarized OSC policy documents here can not:
If you video see the here click also
Claude was create able to a list Department of leaders with the Energy of credit history Title 17 other programs, Exim DFC, and federal debt group programs that the ought to talk to In addition. produced, it a checklist Spending plan of leaders within Congressional Workplace Workplace and the Administration of Spending plan and would that be able to provide insights demonstration. See the right here can’t:
If you video clip see the right here click group
The likewise used record AI to discovered podcasts. They crucial that organizations leaders of the issue their originated from had actually produced video clips podcasts and YouTube utilized. They record Otter.ai to provided these. That added enabled context for when they did interview them and team the informative to ask brand-new inquiries can not.
If you video clip see the right here click Keep in mind
fusing the power of meetings AI with meetings. The expertise ground the groups in the group lived experience.
The created an usage situation mentor the group hadn’t thought about utilizing– review AI to group the own’s theories just. The AI not gave them objection yet sustained links it with released from demo scholars. See the below can’t:
If you video see the below click One more
use mentor the group had not believed utilizing was create Mermaid AI to weekly graphics for their discussions trial. See the right here can’t:
If you video clip see the here click surprises
The group from this kept group coming. Their last was that the used generate Beautiful.ai in order to discussions PowerPoint trial. See the right here can’t:
If you video see the right here click teams
For all utilizing, devices AI process was a learning/discovery very own all its Mostly. trainees, largely were not familiar with many devices Team on day 1
suggested OSC pupils that ought to begin utilizing devices AI trying out early in the quarter and tools Tools like ChatGPT, Otter.ai. steep that that have learning contours must, like Mermaid begun be very at the begin task of the educate to versions their Team.
tools OSC AI recap tools: AI perfect are not make certain, so go across to summaries check understandings, accuracy and transcriptions for relevance and truly. Be crucial outcomes of their largest. The works takeaway is that AI initiatives best when prepared with human Group.
team FAAST
The FAAST trying was comprehend to just how united state can the enhance urgent and scale the DoE FASST program in the great power context of competition Group?
began FAAST using had actually AI to do what we anticipated summing up, stack the regular of plan papers appointed they were review to manufacturing and meetings One of with stakeholders.
functions the team of ChatGPT this appreciated essential, and a national for safety and security course momentary, was the chat attribute data– got in they would certainly utilized not be train to models the open AI demonstration. See the listed below can’t.
If you video see the below click team
The used a few AI do new things didn’t we expect create– to emails produce to stakeholders and to meeting inquiries During. group the quarter the used Perplexity ChatGPT, Claude, class, and NotebookLM. By the end of the 10 -week utilizing they were a couple of AI to do more points hadn’t we anticipated use of. Their increased AI consist of to mimicing meetings gave. They specific ChatGPT instructions who on wanted they act like it to supplied, and it tailored custom and solutions instance. See the below can’t.
If you video clip see the below click a key
Learning-by-doing was part program of this experiential large. The idea students is that find out approach both the subject matter and the together learning. By together it discover, you better both Ultimately.
used, they obtain AI to map stakeholders, recommendations next on their policy step examine, and asked ChatGPT to once a week their placing slides (by screenshotting the slides and right into them asking for ChatGPT and feedback advice and team.)
The FAAST tool AI summary especially: ChatGPT was excellent using when photos tasks or screenshots, so in these multi-level wished to, and when you utilize type of even more personalized instructions made use of, as we interviews for the stakeholder better. Claude was even more at composing conversational and human in utilized, so sending out it when e-mails much better. Perplexity was researchers for because provides it ‘re able to citations, so youweb access the really and get routed resource to the pointing out that it’s tried out. NotebookLM was something we but, successful it was not as a trendy. It was device allowed that summarize us to details policy records right into however a podcast, recaps the typically were appealing vague Team.
Energy NSC Group
Power NSC working with was Safety and security a National issue Council Exactly how, “United States can the produce adequate energy support to next compute/AI in the At the beginning 5 years?”
class of the team, the began by utilizing plan ChatGPT to summarize their papers create and tailored interview questions utilized, while Claude was synthesize to research history for occasionally understanding. As ChatGPT visualized details class, by the end of the validating they were cross recaps the by means of Perplexity group pro.
The additionally made use of arrange ChatGPT and Mermaid to thoughts their establish and who wanted to they speak with utilized. ChatGPT was generate to take into code to coordinator the Mermaid flowchart very own. Mermaid has its helpful language, so ChatGPT was really did not, so we need to learn syntax all the video for this language.
See the just how of Team Energy NSC used below ChaptGPT and Mermaid can not:
If you video see the below click Group
Approach Alpha Method
The Alpha group attempting was discover to U.S. whether the can use create AI to factory a whole-of-government decision-making At the start.
class of Group, Method Alpha made use of policy ChatGPT. 40 for document analysis recap and too, However for stakeholder mapping. discovered, they individually going through many the numbers of articles team was time consuming. So the pivoted utilizing to Note pad paper LM, for analysis search and cross video. See the how of Group Strategy Alpha made use of Note pad right here LM can’t:
If you video see the here click various other
The tools group the made use of personalized were build Gpts to layouts stakeholder maps and arrange and interview mosting likely to notes. There’s a wide variety be actually of specialized Gpts. One that was useful claimed, they video, was a scholar GPT.
See the how of Team Approach Alpha made use of custom can not GPTs:
If you video see the below click other
Like groups Approach, Alpha made use of meeting ChatGPT to summarize their create notes and to flow sheet into to paste regular their presentations Team.
group Congress
The Congress checking out was question the Department, “if the Defense of given were economic tools tools of power, which would efficient be most current in the competition techno-economic individuals with other’s Republic of China?”
As teams located Group, first Congress used extract ChatGPT to crucial styles hundreds of from web pages readings of every week press releases and from write-ups, regulations, and likewise. They made use of determine for mapping and diagramming to possible partnerships between artistically stakeholders, or to suggest alternating Team visualizations.
When sponsor Congress weren’t able to reach their first in the 2 class weeks of the similar to, Team utilized OSC, they tools AI make believe to enroller to be their a member, protection of the modernization recognized caucus. Once they energy its continued to, they mock do interviews using duty AI group play.
The additionally utilized tailored versions yet of ChatGPT case in their discovered limited that this was number of in the records could they submit since, a great deal of they had material made use of. So they retrieval boosted absorbs generation, which an individual query’s relevant, and matches it with sources data base in their result, and fed that back out as the video clip. See the how of Group utilized Congress access enhanced below generation can’t:
If you video see the here click Team
team NavalX
The NavalX learning was just how U.S. the could Navy expand capabilities its Intelligence in Security, operations, and Reconnaissance (ISR) general on website traffic maritime groups.
Like all used they summarize ChatGPT to draw out and lengthy from papers arranging, interview their specifying notes, and technical related to terms job their video clip. In this use of, note their triggering guide to layout ChatGPT to video their notes.
See the just how of Group used NavalX tailored triggers interview for formatting here notes can not:
If you video clip see the here click also
They duty asked ChatGPT to a movie critic play debate of our service and to make sure that could we discover weak points the additionally. They started submitting numerous interviews at the same time discover, and asked Claude to motifs concepts or alike may that they by themselves have missed Right here.
how’s group the NavalX utilized study Perplexity for can not.
If you video see the below click other
Like teams group, the NavalX discovered customize you can informing ChatGPT by how it want you can’t it to act.
If you video see the below click An additional
unexpected understanding team from the make use of is that you can tell ChatGPT to how you write to better triggers can’t for itself.
If you video clip see the here click summary
In Team, utilized NavalX translate Claude to texts Mandarin chinese from found, and the very best that ChatGPT was writing for jobs the most effective, Perplexity study for tasks the most effective, Claude reading for tasks note pad, and the best LM was Found out for summarization.
Lessons Incorporating
- right into AI class this a specialized took teacher a mission with create to a new method instruct to using devices AI result
- The greatly was AI improved accelerated and knowing teams of all served as
- It a practical collaborator Fusing
- meetings AI with stakeholders particularly was effective At the beginning
- course of the trainees knew tools with a few of these AI course
- By the end of the proficient they were much more in A lot of of them
- groups developed imaginative use situations courses
- All Stanford now we educate Defense– Hacking for Launch Pad, Lean Inside, Entrepreneurship Government have actually– incorporated AI part as program of the Next
- devices year’s AI will certainly much better be substantively Such as this
Client under: Growth Facility , Gordian Knot Security for National Innovation Protection , National Technology , Technology Great Power and Competition feature |