🌿 THE GOOD AI
The industries most exposed to AI just hired more people and paid them more
The single loudest fear about AI is that exposure to it means fewer jobs and lower wages. A peer-reviewed study published in April 2026 looked at what actually happened between 2017 and 2024 and found the opposite. Industries with the highest exposure to generative AI added jobs faster and raised wages faster than less-exposed sectors. Per standard deviation of AI exposure, the researchers measured a 10% productivity increase, 3.9% job growth, and 4.8% wage growth.
The mechanism matters as much as the numbers. Sectors where AI complements human tasks saw employment rise. Sectors using AI more autonomously showed no significant job loss either. That is a different picture than the one a lot of 2023 and 2024 coverage promised, and it holds up because it is drawn from seven years of real labor data, not a survey of executive intentions.
There are caveats worth naming. This study looks backward through a window that mostly predates the most capable agentic systems. The gains have been uneven across roles, and the study averages across a lot of variation underneath. Wage growth tracked productivity but did not match it one for one, and we still have little to say about which specific jobs moved and which did not.
What it does establish, rigorously, is that AI exposure and employment growth can run in the same direction. That is a result, not a forecast. It should push the conversation away from "will AI take our jobs" and toward "what do workers in exposed sectors need to keep capturing these gains."
Source: Phys.org
⚡ 3 GOOD SIGNALS
⚡ Healthcare AI is no longer experimental, 70% of organizations now run it in production
NVIDIA's second annual State of AI in Healthcare and Life Sciences survey of 600-plus professionals found 70% of healthcare organizations actively using AI, up from 63% in 2025, with generative AI adoption jumping from 54% to 69% year over year. The ROI is landing too. 57% of medtech respondents report measurable returns in radiology and imaging, 85% say AI is increasing revenue, and 80% report cost reductions. 85% of executives plan to spend more this year.
Source: NVIDIA Blog
🤝 Lenovo launches an AI lab for nonprofits, with hardware and engineers attached
Lenovo and Tech To The Rescue have launched the AI for Social Impact Lab, equipping 12 climate and education nonprofits with Lenovo hardware, AI tools, and matched volunteer technologists to design and deploy real solutions. The lab sits inside the broader AI for Changemakers Accelerator, backed by AWS, Google.org, and others, aiming to lift 110 nonprofits. Access to compute and technical support is the two-part bottleneck nonprofits most often cite. This program targets both.
Source: Lenovo News
🌱 AI's power hunger is accidentally reviving carbon capture at scale
At least five US projects are now exploring carbon capture on natural gas plants that feed AI data centers, Axios reported April 13. Google's Illinois project is designed to capture about 90% of CO2 from a 400-megawatt facility. Meta and projects involving ExxonMobil and Chevron are also in play. AI's energy appetite is a real concern. The unexpected upside is that it is finally creating the economic case for carbon capture the technology never had before.
Source: Axios
🔬 THE DEEPER DIVE
A new Stanford SIEPR study tracking 200,000 US households found that generative AI users are 76% to 176% more efficient at everyday "digital chores": job hunting, travel planning, shopping, paying bills, and navigating government websites. That is an enormous time dividend for individuals, and it is showing up in the household data the same way productivity gains are starting to show up at work. Cross-referenced with Stanford HAI's 2026 AI Index Report, released in mid-April, the picture sharpens further. Generative AI has reached 53% population adoption within three years, outpacing the PC and the internet, and the median value per user roughly tripled between 2025 and 2026.
The catch is in the distribution. Younger and higher-income users are adopting GenAI tools substantially faster than older and lower-income users. The Stanford authors called it a "GenAI digital divide" and warned that it is widening, not closing.
The PM lens
For product managers, the useful signal is that consumer AI has crossed from novelty to utility for a specific demographic, and the next wave of growth depends on reaching the users who are not there yet. The 76% productivity finding is a demand-side data point. It says the value is real and measurable, which changes how to price, position, and onboard. The wider the feature set, the higher the cognitive overhead for new users, and the smaller the share of the population that will actually capture the gains. Product surfaces designed for AI-fluent power users will keep widening the divide. Product surfaces designed for first-time users will narrow it, and probably win the bigger market over time.
The risk lens
From a risk perspective, the equity story is where responsible AI stops being a governance abstraction and starts being a measurable outcome. If AI is delivering a 76% to 176% personal productivity lift, and if access is skewed toward people who already have time, money, and digital skills, the compounding effect runs the wrong way fast. The Stanford data gives policymakers, educators, and employers a number to work against. Libraries, community colleges, and benefits agencies are natural distribution points for the kind of access programs that could narrow the divide before it hardens. The downside risk is not just unfair outcomes, it is a political backlash that treats AI as a privilege technology and undermines the broader optimistic case.
The next 12 to 24 months
Expect more datasets like this one, measuring consumer AI value the way we already measure workforce AI impact. Expect the policy conversation around AI access to sharpen, moving from "digital literacy" language to specific, funded access programs. The optimistic scenario is simple. The productivity lift is real, the adoption curve is fast, and the only question is who gets to ride it.
Sources: Stanford News , Stanford HAI
🛠 TOOL OF THE WEEK
Abridge
Abridge is an ambient AI scribe that listens to the clinical conversation between doctor and patient and generates a structured clinical note in real time. It is deployed across Intermountain Health, UCSF, Kaiser Permanente, and other major US health systems, and it sits behind the results the American Hospital Association just profiled. Across a new JAMA study of five academic medical centers, ambient scribes cut EHR time by 13.4 minutes and documentation time by 16.0 minutes per session, enough for clinicians to see one additional patient every two weeks. 94.7% of AI-generated notes were free from significant errors. This is AI handling the part of the job that was eating doctors alive.
→ Read more: American Hospital Association
💬 ONE QUESTION
If AI is already giving people a 76% productivity lift at home, which of your own everyday tasks have you quietly handed off to it, and which ones do you still insist on doing yourself?
Hit reply. We read every response.