New Oxford and Google DeepMind paper finds most workers value jobs beyond pay, so basic income alone may not offset AI job losses. So plans for the AI transition should protect purpose, community, and dignity at work as well as income. finds that about 75% of workers would quit if guaranteed 100% of their wage, yet 1 in 5 would refuse even 120%. If AI automates many jobs, a common plan is a basic income that replaces lost wages. That plan works only if people work mainly for the money. The author surveyed 8,234 working-age adults in the UK, US, Denmark, and India. Workers were sorted by what they value most in their job and asked what guaranteed income would make them quit. Only 26% of classified workers valued their job mainly for pay, while one third valued the work itself. 20% would keep working even if offered 120% of their salary as guaranteed income, a share similar across income levels.

In our work, we evaluate 11 economic policies for AGI across four dimensions—welfare, agency & empowerment, implementation feasibility, and durability across Mild, Moderate, and Full Transformation scenarios—using an exploratory mixed methods approach.

Very excited about the release of the ATLAS 1.0 white paper on AI use and the economy. The report is a broad collaboration between @Google & @GoogleDeepMind and covers how our AI is being used at scale. The data set has 15 million de-identified interactions across a large number of surfaces including the Gemini App, API, and Google's AI mode. This broad coverage gives us perspective on how AI is used both at work and outside of work. The team did a huge amount of work classifying interactions to specific tasks at work and at home. There are tons of interesting findings in the paper (link below) but here are some of my favorites: 1. Yes, white collar work is overrepresented in work-related AI usage, but it's also being used for a non-white collar work. It's used as a hands on collaborator for diagnostics and troubleshooting. Manual and technical trades tend to use the multimodal features more, images and video, e.g., auto techs and industrial mechanics interpreting complex test results, inspecting machinery for wear, debugging electrical wiring. 2. There is super broad diffusion of AI for almost everything. AI use covers occupations that represent 88% of total US employment, white collar work but also farmers and foresters. 3. AI is creating real value outside of work that GDP statistics may miss. This supports @erikbryn's proposal that GDP will be an incomplete measure of AI's positive impact on society. A huge number of AI interactions are in "productive household activities" such as researching things for the home, interior design, help with tools/appliances etc. Interactions regarding legal services are also highly overrepresented. 4. One of the most overrepresented uses of AI for non-work activities is interacting with the public sector. Activities related to interactions with govt services and civic obligations make up a huge number of interactions, and importantly, many take place outside of business hours. 5. Both #3 and #4 suggest to me that AI will be very useful in helping people navigate complexities and overcoming frictions/barriers in their every day lives. So many conversations about AI focus on labor market impacts, but welfare is not just work, and ATLAS nicely highlights that the broader impact of AI is quite significant. 6. We cover a good deal of global diffusion. English accounts for only 1/3 of global conversations and users engage with AI in their native language for both complex work and non-work tasks. Interestingly, non-OECD countries have a lot more image and video usage. The reason it's called 1.0 is because this is only the beginning. We have a ton of follow up projects in the works. ai.google/static/documents/G…

Excited to see this paper out from the excellent @JulianDJacobs, part of our economics group at @GoogleDeepMind, and Jordan Canedy. They look at the impact of prior worker retraining programs, finding that contrary to their purpose, most workers stay in the same or similarly exposed sector after retraining. This is the first paper from our group. Much more to come!

What if worker training programs like those we might think of to address #AI-driven dislocation don't work? My friend @JulianDJacobs shows the problem @alexolegimas @GoogleDeepMind @JustinHeck @sebkrier @sj_manning

The main federally-funded worker retraining program in the US (WIOA) rarely shifts workers out of automation-vulnerable work: 45% of participants return to their prior industry and 27% stay in the same occupation. The authors use RTI (how routine/automatable a job's tasks are), which proxies the computerization wave of automation workers faced over 2017–2023. They (rightly) avoids 'AI exposure' as a metric since exposure doesn't tell you whether AI complements a worker (more productive, better paid etc) or displaces them. arxiv.org/abs/2605.03767

Authors: @weballergy @jzl86 @JulianDJacobs @sindero Read here: papers.ssrn.com/sol3/papers.…

Job retraining doesn't work, and this is for a number of reasons: challenges for participants (costs, lack of interest), providers (bureaucracy, inability to match skills to a changing market), and a fundamental lack of sufficient high-quality jobs. aipolicyperspectives.com/p/a…

Does anybody have recommendations for an independent freelancer who could take an LLM and build an application to summarise a number of different newsletters, in a way that I could refine over time? Ideal skillset would be: familiar with and interested in the latest models; strong coding expertise; skilled at packaging up content for a time-poor audience + discerning audience. Feel free to comment under this post if this is you or you know anyone interested - not sure I'll do this yet but want to see who might be able to do so.

And always the Mets.

I created a median house cost vs. median household income chart by state/region for the US/UK. It really highlights the scale of the problem - even "affordable" regions of the UK are comparable to some of the least affordable states, and London is vastly worse than any state.

.@MarkMuro1 & @JulianDJacobs write that there are several AI research programs across the U.S. showing how federal initiatives can counter excessive geographic unevenness and broaden the technology’s development. Read more: brookings.edu/articles/the-c…

Regional tech ecosystems are essential but the clustering of #AI design work into just a few core metros isn't healthy Here's my new post with @JulianDJacobs on why and what to do brookings.edu/interactive/ge… @BrookingsInst @BrookingsMetro brookings.edu/interactive/ge…

The case for promoting the geographic and social diffusion of #AI development | My latest @BrookingsInst with @JulianDJacobs brookings.edu/articles/the-c… @BrookingsMetro

#AI creation is hugely concentrated in just a few metros. Here's why we should actively encourage its diffusion--and how we can do that. My new post with @JulianDJacobs brookings.edu/interactive/ge… @BrookingsInst @BrookingsMetro @robseamans @sj_manning @erikbryn @I_Am_NickBloom

Technology-driven automation has created economic insecurity that is partly responsible for rising populism and political polarization across the U.S. and Europe. This populism tends to favor the far right, @JulianDJacobs and co-authors emphasize. brookings.edu/articles/does-…

Five ways communities can sustain place-based industrial initiatives beyond initial awards: --Follow-on federal $ --State and local $ --Philanthropy --University support -Corporate $ See our new piece here: brookings.edu/articles/finan… @BrookingsInst @BrookingsMetro @dgoetzel