AI at Work
AI for DC
Hi
Welcome (back) to The Prompt. AI isn’t well understood, but we learn a lot in our work that can help. In this newsletter, we share some of these learnings with you. If you find them helpful, make sure you’re signed up for the next issue.
[Insight] AI at work
No technology in history has spread faster than AI. ChatGPT, launched less than three years ago, is now used weekly by 800 million people—about 10% of the global population. The speed and scale of AI adoption have fueled competing narratives about AI’s potential impacts on the workforce. Some predict widespread job losses; others an imminent productivity boom.
Like many polarized debates, the truth lies in between. New jobs will be created; others will evolve; some jobs will disappear—we should be clear-eyed about that. To help prepare for this shift and ensure that AI’s benefits reach everyone—not just the few—we’ve released a new Workforce Blueprint. It’s a starting point for policymakers, businesses of all sizes, and labor to work together to begin building this new opportunity for an AI-powered economy.
For now, the data about AI’s economic impact tell a nuanced story. OpenAI recently analyzed ChatGPT usage, the largest dataset in existence on actual use of AI, from three different angles to better understand who is using generative AI, what they’re using it for, and what it can do in today’s economy.
Interpreting these results requires humility. We live in an incredibly dynamic era and our tools are evolving fast. But studies like these point to an emerging pattern: so far, AI is more enabler than replacer.
On consumer use, our analysis of 1.5 million de-identified ChatGPT conversations found that three kinds of activity dominate: seeking information, getting practical guidance, and writing, together accounting for nearly three-quarters of all user interactions. About one-third of these messages on the consumer platform appear to be work-related. This kind of use doesn’t fully register in GDP, but it creates what economists call consumer surplus by saving time and improving decisions. One recent independent estimate by economists Avinash Collis and Erik Brynjolfsson measured this value at nearly $100 billion per year for consumers, most of whom are using ChatGPT for free.
At the workplace, ChatGPT adoption follows a bottom-up pattern. Employees often begin using ChatGPT at work before their employers introduce it. Within the first 90 days of companies deploying ChatGPT, workers use it for core business tasks like researching markets and data analysis, with writing consistently leading across all functions, from finance to sales and legal. Similar to consumer use, AI’s greatest value at work lies in improving how people write, reason, and make decisions. People are using AI to help them make decisions and streamline routine chores and duties. In short: more ask than task. These gains matter economically, though not in a way that traditional indicators capture.
On the timing of the transformation, OpenAI researchers developed a new evaluation, called GDPval, which measures how well models perform on economically valuable tasks sourced from real-world practitioners, from writing a legal brief to conducting market research. The results: GPT-5-level systems now match or exceed human professionals on about half these tasks, completing them in minutes instead of hours. If current trends hold, these capabilities will advance rapidly over the next few years. – Ronnie Chatterji, OpenAI’s Chief Economist
[Policy] Proposals we’re already building
Among the proposals in our Workforce Blueprint—from programs OpenAI already has built or that we’re about to build:
Match workers with new opportunities. That’s why we’re building the OpenAI Jobs Platform, to help match knowledgeable, experienced candidates at every level with big companies, small businesses, and local governments that are looking for AI-skilled talent.
Make AI certifications and training broadly available. One pathway we’ve already created is the OpenAI Academy, a free online learning platform that connects AI learners with the resources, workshops, and communities they need to master AI tools. Another pathway is OpenAI Certifications, which will certify employees of companies large and small, state and local government workers, start-ups and individual entrepreneurs in multiple levels of use of ChatGPT.
Develop training tools that teach, not just answer. OpenAI built StudyMode as a learning experience that helps users work through problems step by step instead of just giving them an answer. Workers could use tools like StudyMode to shorten the time it takes to learn the new skills required for a promotion, to develop new subject matter expertise that allows them to pivot into a different career field, or to study for professional exams and certifications.
Build cross-sector partnerships for the Intelligence Age. To ensure that the Stargate AI infrastructure investment translates directly into local AI literacy and opportunity, we propose connecting Stargate datacenter campuses with community workforce and education partnerships. By working with community colleges, technical schools, and other community partners, this approach will ensure that in every community where we build, we are also building pathways for people to thrive in the AI-powered economy.
These are just a few of the steps we can take to help people thrive in an AI-powered economy. Read the entire Blueprint here.
[Prompt] The jobs that AI can create
Last month, we published a report about the new kinds of job roles that the Intelligence Age can produce. We also asked ChatGPT for its thoughts on the new jobs that AI can create. Check it out.
[About] OpenAI Forum
On Monday, October 20, the OpenAI Forum is hosting its inaugural Higher Education Guild – to showcase use cases of AI in teaching and academic research. This invite-only event is taking place in San Francisco.
5:00 PM – 10:00 PM EDT on Oct 20
[About] OpenAI Academy
The Academy, which is celebrating its one-year anniversary, is OpenAI’s free online and in-person AI literacy trainings for beginners through experts.
OpenAI has called for a nationwide AI education strategy – rooted in local communities in partnership with American companies – to help our current workforce and students become AI-ready, bolster the economy, and secure America’s continued leadership on innovation.
2:30 PM – 3:30 PM EDT on Oct 30
[Disclosure]
Graphics created by Base Three using ChatGPT.
This strikes the right balance, acknowledging both disruption and opportunity. The “more ask than task” framing perfectly captures how AI is reshaping work by amplifying human reasoning rather than replacing it.
This is a really insightful breakdown, especially the distinction between AI as an 'enabler' versus a 'replacer' so far. The data showing usage leans heavily toward seeking information, guidance, and writing support—'more ask than task'—is super telling about its current utility!
I especially appreciate the proactive steps outlined in the Workforce Blueprint. Given how incredibly fast AI evolves, offering concrete training pathways like the OpenAI Academy and the new Certifications feels crucial. It's hard enough to keep up with the tools, let alone integrate them meaningfully into work, so these initiatives look like they'll really help bridge that knowledge gap for a lot of people trying to catch up. Great starting point!