The risks and the possibilities of AI, side by side, with every number traced to its source.Move across the picture and the light follows.
Most plants require both light and dark to meet their furthest growth potential.
So it is with us.
Some people look at AI and feel dread. Some feel hope. Most of us feel both on the same day. Here the evidence for each sits next to the other, so you can carry them together.
I use AI in my work and in my art. I also see the anger and the fear around it, and how it has split people who make things into camps. That division has weighed on me.
I do not see AI as good or bad in itself. There are serious concerns, and there is also real good already growing. I wanted one place where the two could sit together, with every number checked against its source.
Michael Hummel
Think of a tree cut across. Each ring is a year the tree lived through, good and hard alike. These nine rings are the places AI reaches into a human life.
Every area holds a wonder and a worry. Each one has many more aspects than the few shown here.
Every figure on this site names its source and its kind. A survey tells us what people believe. It cannot tell us how likely an event is. So each number carries one of five labels:
Press a small number like this one to see where a figure comes from.
Wonder
consent that works
30,000+
artists take part in one opt-in licensing program that now shares 25% of net revenue with them. Reported
Worry
consent, honesty, likeness
73%
of adults in an 8-country survey say it is unethical for AI firms to use copyrighted music without the artist's approval. Survey
Wonder
shared prosperity
11 more hours
a week of creative work reported by artists in Ireland's basic income pilot than by a lottery control group. Measured
Worry
livelihoods
26% and 36%
of illustrators and translators who answered a UK survey said they had already lost work to generative AI. Survey
Wonder
a better-watched planet
97.2%
of 1,320 forecast targets on which an AI weather model beat the leading conventional ensemble forecast. Measured
Worry
energy and water
1.5%
of the world's electricity went to data centers in 2024. Use is projected to more than double by 2030. Estimate and forecast
Wonder
a patient teacher for all
Over double
the median test gains with a well-designed AI tutor, against an active-learning class, in a randomized trial of 194 Harvard physics students. Measured
Worry
learning and skill
+48%, then 17% worse
Students using plain GPT-4 for math practice scored 48% better in practice, then 17% worse than classmates on an exam without AI. A tutor with guardrails largely avoided the harm. Measured
Wonder
connection and care
19.6%
higher on an automated empathy measure: peer supporters' replies written with AI feedback. It was 38.9% for those who said support was hard. Measured
Worry
relationships and mental health
72%
of US teens have tried an AI companion, and 52% use one at least a few times a month. Survey
Wonder
truth and common ground
About 20%
drop in belief in a chosen conspiracy theory after a conversation with an AI, among 2,190 believers. It lasted at least two months. Measured
Worry
truth and trust
55.5%
is how often people correctly identify deepfakes, which is not reliably better than chance (50%). Measured
Wonder
access for everyone
200 languages
translated by one model released free for non-commercial use, many with few written resources. Measured
Worry
who bears the cost
27.5% vs 15.4%
of working-age people used generative AI in the Global North vs the Global South in early 2026, and the gap is widening. Estimate
Wonder
more voices, new forms
10.7%
more novel: stories by less-creative writers who were offered up to five AI ideas. Measured
Worry
sameness and flooding
Up to 8%
more novel is how stories were rated when writers were given AI ideas. The stories also became measurably more alike. Measured
Wonder
longer lives, a wiser hand
29% more
cancers found by AI-supported mammography (6.4 vs 5.0 per 1,000) in a randomized trial of 105,934 women, with no rise in false positives and 44% less reading work. Measured
Worry
control, power, meaning
5%
is AI researchers' median guess at the chance that AI causes human extinction or severe disempowerment (2023, 15% response rate). Survey
You can be worried about where this is going and still be moved by what it can do. Carrying both is harder than picking a side... and it leaves more room for other people.
Eight commitments that an artist who refuses AI and an artist who uses it could both sign.
My own everyday guidelines, the Shared Ethic for art and music, and what thirteen fields now ask of their own people, from medicine and law to faith traditions. Underneath the differences runs a common ground: a person stays accountable, checks the output, tells the people affected and protects their privacy.
MedicineMental healthLaw and the courtsJournalism and mediaEducationResearch and publishingArt, music and writingBusiness, HR and financeGovernment and regulationTechnology buildersChildren and familiesFaith and philosophyEnvironment and energy
Starting 9 November 2026 Published
A sham-controlled study in a small village of AI agents asks whether a computed inner "heartbeat" changes what they do, beyond the true facts it carries.
Starts 9 November 2026. Results, and what they might mean, will be added in February 2027.
Field note
Can a machine have something that works like a feeling? A sham-controlled study in a small village of AI agents begins on 9 November 2026. The plan, the reasoning and the ethics are published now. Results are due in February 2027.
New page
A new Guidelines section brings together my everyday guidelines, the Shared Ethic for art and music, and a table of the common ground that runs through every profession's rules.
New page
New pages on how medicine, mental health, law, journalism, education, research, the arts, business, government, technology builders, children's services, faith traditions and the environmental field are setting rules for AI, each with its sources and questions to ask yourself.
Source 41
King A. LANDR Fair Trade AI program. Digital Music News, 14 Jul 2026. digitalmusicnews.com/2026/07/14/landr-fair-trade-ai-program-indie-artists/ ; MusicTech, 13 Jul 2026. musictech.com/news/music/landr-fair-trade-ai-advance-program/
Source 1
Deezer/Ipsos. Survey of 9,000 adults in 8 countries on AI music. Deezer Newsroom, 12 Nov 2025. newsroom-deezer.com/2025/11/deezer-ipsos-survey-ai-music/
Source 43
Department of Culture, Communications and Sport (Ireland). Basic Income for the Arts: two-year report, and Alma Economics cost-benefit analysis. gov.ie, Sep 2025.
Source 4
Society of Authors. AI survey results (n=787, Jan 2024), via European Writers' Council. europeanwriterscouncil.eu/soa-survey-uk-ai-2024/
Source 45
Price I, et al. Probabilistic weather forecasting with machine learning. Nature 637:84-90, 2025. doi:10.1038/s41586-024-08252-9
Source 8
International Energy Agency. Energy and AI. Apr 2025. iea.org/reports/energy-and-ai/executive-summary
Source 48
Kestin G, et al. AI tutoring outperforms in-class active learning. Scientific Reports 15:17458, 2025. doi:10.1038/s41598-025-97652-6
Source 13
Bastani H, et al. Generative AI without guardrails can harm learning. PNAS 122(26), 2025. doi:10.1073/pnas.2422633122
Source 50
Sharma A, et al. Human-AI collaboration enables more empathic conversations in text-based peer-to-peer mental health support. Nature Machine Intelligence 5:46-57, 2023. doi:10.1038/s42256-022-00593-2 (300 peer supporters, non-clinical)
Source 16
Common Sense Media. Talk, Trust, and Trade-Offs: How and Why Teens Use AI Companions. Jul 2025 (n=1,060). commonsensemedia.org/research/talk-trust-and-trade-offs-how-and-why-teens-use-ai-companions
Source 52
Costello TH, Pennycook G, Rand DG. Durably reducing conspiracy beliefs through dialogues with AI. Science 385(6714), 2024. doi:10.1126/science.adq1814
Source 19
Diel A, et al. Human performance in detecting deepfakes: a systematic review and meta-analysis of 56 papers. Comput Hum Behav Rep 16:100538, 2024. doi:10.1016/j.chbr.2024.100538
Source 54
NLLB Team. Scaling neural machine translation to 200 languages. Nature 630:841-846, 2024. doi:10.1038/s41586-024-07335-x
Source 22
Microsoft AI Economy Institute. Global AI Diffusion, Q1 2026. May 2026. microsoft.com/en-us/research/wp-content/uploads/2026/05/Microsoft-AI-Diffusion-Report-2026-Q1.pdf
Source 26
Doshi AR, Hauser OP. Generative AI enhances individual creativity but reduces the collective diversity of novel content. Science Advances 10(28), 2024. doi:10.1126/sciadv.adn5290
Source 58
Hernström V, et al. Screening performance and characteristics of breast cancer detected in the MASAI trial. Lancet Digital Health 7(3):e175-e183, 2025. doi:10.1016/S2589-7500(24)00267-X
Source 59
Gommers J, et al. MASAI trial, interval cancer results. Lancet 407:505-514, 2026. doi:10.1016/S0140-6736(25)02464-X
Source 28
Grace K, et al. Thousands of AI Authors on the Future of AI. J Artif Intell Res 84, 2025. arxiv.org/abs/2401.02843
