简体中文
English
Website review
This review may include AI-generated content and is not individually verified. Domain, certificate and other facts come from the observed data on this page.
What is It Does What Now?
It Does What Now? is a chronological news timeline of artificial intelligence from January 2020 onward, published at itdoeswhatnow.com. Instead of explaining how AI works, it records what happened: model releases, funding, policy fights, court cases, and accidents, each as a short brief with links to the original reporting. The homepage headline frames it as "What's happening right now with AI — and how we got here."
Its organising idea is that the story is easier to follow in order than as a stream of separate headlines. Entries are grouped by year (2020 through 2026) and tagged by track, so you can read straight through or filter down to one thread. The tracks listed on the page include Models & capabilities, Benchmarks & progress, Labs & people, Safety & alignment, Security & misuse, Government & policy, Courts & copyright, Money & business, Compute & infrastructure, Ideas & essays, Open weights & ecosystem, and Culture & impact.
Who it suits
- Someone who wants the background to a current AI story without reading a dozen separate explainers.
- Researchers, students, or writers assembling a dated sequence of events who need a starting point for sources.
- Anyone who has been ignoring AI news for a year and wants a catch-up rather than a tutorial.
It is less useful if you want technical depth, hands-on reviews, or a single opinionated argument. The briefs are described as neutral summaries, and the site separates that from a commentary section, where outside writers' takes are collected under their own names — for example items attributed to Don't Worry About the Vase, Transformer, and Epoch AI. That split is the most useful thing about the format: facts in one column, argument in another, so you can tell which you are reading.
A practical way to use it
Suppose a headline says a company pulled a model for safety reasons. Open the timeline at that date, read the brief, follow its source links, then step back a year in the same track — Safety & alignment or Labs & people — to see whether this is a new pattern or a recurring one. The filtering is what makes that possible; a plain reverse-chronological blog would not.
For the wider landscape, a general technology news outlet such as Ars Technica covers AI daily, and MIT Technology Review runs longer analysis. Use this timeline for sequence and sourcing, and those for depth.
Note the pricing signal on the page: the only subscription link points to a newsletter hosted on Substack, so the timeline itself appears readable without an account.
How can I use It Does What Now? to track recent AI developments?
It Does What Now? works as a reverse-chronological news log for AI, so the fastest way to use it is as a timeline you scan rather than a site you read front to back.
Start with the newest entries. The homepage opens on the current year and lists dated briefs, newest first — for example, 29 September items on Anthropic finding a Chinese open-weight model capable of building working cyber exploits, China reporting over 700 million generative AI users, and OpenAI launching always-on personal agents at DevDay. Each item is a short neutral summary with links to its original sources, so you can skim headlines and open only the ones that matter to you.
Then use the filters to narrow by interest. The page exposes a set of tracks you can filter by, which is the site's main advantage over a general news feed:
- Models & capabilities
- Benchmarks & progress
- Labs & people
- Safety & alignment
- Security & misuse
- Government & policy
- Courts & copyright
- Money & business
- Compute & infrastructure
- Ideas & essays
- Open weights & ecosystem
- Culture & impact
You can also adjust how much detail appears — the page offers Headlines, Major, Notable and Everything views — and jump to a specific year from 2020 through 2026.
Use the commentary section for the argument, not the event. A separate stream collects outside writing and opinion from across the period, with named sources and dates. That is where you go when you already know what happened and want to know how people are reacting.
A practical routine: check the Headlines view weekly for events, switch to a single track when a topic becomes relevant to your work, and use the year links when you need to reconstruct how a debate started. If you want a second timeline to compare coverage against, Epoch AI publishes research on AI trends and compute, and arXiv is where many of the underlying papers the briefs cite actually live.
What are the key AI events covered from 2020 to now?
The site is a chronological AI timeline running from January 2020 through September 2026, organised by year and tagged by track. Its value is breadth plus ordering: instead of a single narrative, it logs the models, funding, policy fights, lawsuits and accidents as dated briefs, each linked to an original source. That makes it useful for two different readers — someone catching up on the last few years, and someone checking what happened in a specific month.
H3 What the timeline actually tracks
The filter list shows the scope: Models & capabilities, Benchmarks & progress, Labs & people, Safety & alignment, Security & misuse, Government & policy, Courts & copyright, Money & business, Compute & infrastructure, Ideas & essays, Open weights & ecosystem, and Culture & impact. A separate commentary stream collects outside voices, such as It Does What Now?'s own timeline entries sitting alongside posts from outlets like Don't Worry About the Vase, Transformer and Epoch AI.
H3 Representative entries
From the page, the recent end of the timeline includes:
- Anthropic finding that a Chinese open-weight model could build working cyber exploits.
- China reporting more than 700 million generative AI users.
- OpenAI launching always-on personal agents at DevDay, alongside a cheaper model tier.
- A US executive order changing official terminology to "super intelligence" and launching a government chatbot portal.
- AI firms and building unions forming an alliance on data-centre standards.
- AMD agreeing to buy Fei-Fei Li's World Labs.
Earlier years follow the same pattern: scaling papers and early model releases in 2020–2021, the generative-AI breakout and copyright fights in 2022–2023, then regulation, compute buildout and safety arguments through 2024–2026.
H3 How to use it
Pick your entry point by purpose. If you want the arc, read the year-in-brief summaries in order. If you are checking a claim or writing something, use the track filters to isolate one thread — policy, say, or compute — and follow it across years without wading through model releases. The headline/detail toggle lets you skim first and expand only what matters.
One caveat worth knowing before you rely on it: dated news briefs reflect the framing of their moment, and fast-moving stories get superseded. Treat each entry as a pointer to its linked source rather than a settled account, especially for anything from the current month.
How does It Does What Now? organize AI news by categories like models, money, and policy?
It Does What Now? organizes AI news as a single dated timeline, then lets readers slice that timeline by track. The homepage shows the newest briefs first, each a short neutral entry with links to original sources, and the same archive extends back to January 2020.
How the categories work
Each entry is tagged to one or more tracks. The page lists these tracks with running counts, so you can see where the archive is heaviest:
| Track | What it covers | Count shown |
|---|---|---|
| Models & capabilities | New models and what they can do | 524 |
| Benchmarks & progress | Measured gains and evaluations | 161 |
| Labs & people | Companies, researchers, org changes | 106 |
| Safety & alignment | Alignment work and concerns | 267 |
| Security & misuse | Exploits, abuse, security incidents | 181 |
| Government & policy | Regulation, agencies, official action | 232 |
| Courts & copyright | Lawsuits and legal rulings | 126 |
| Money & business | Funding, deals, acquisitions | 268 |
| Compute & infrastructure | Chips, data centres, capacity | 154 |
| Ideas & essays | Commentary and analysis | 127 |
| Open weights & ecosystem | Open models and tooling | 212 |
| Culture & impact | How AI lands socially | 196 |
The counts are a useful signal in themselves: models and capabilities dominate, money and business is close behind, and courts and copyright is the smallest major track — so a reader interested in law will find a thinner but more focused stream.
Practical ways to use it
- Follow one thread over time. Filter to Government & policy, then read down the years to see how regulation moved from proposals to orders.
- Check the depth setting. Detail levels range from Headlines through Major and Notable to Everything, so you can skim a week or read the full record.
- Jump by year. Year sections (2020 through 2026) let you start at a period rather than scrolling the whole archive.
- Read the commentary separately. Opinion and essays sit alongside the neutral briefs, attributed to their authors, so you can tell reporting from argument.
A concrete scenario: you are preparing a briefing on AI security and want recent examples rather than a general overview. Filter to Security & misuse, set detail to Major, and read the last few months — the 29 September entry on Anthropic finding a Chinese open-weight model capable of building working cyber exploits is exactly the kind of item that track is for. Then cross-check the same date under Government & policy to see the official response.
Next step: pick one track that matches your interest, set the detail level to Major, and scan a single year before widening out. If you want a second dated source to compare framing, the timeline format is also used by Epoch AI, whose research notes pair well with the benchmarks and compute tracks here.
What sources does It Does What Now? use for its AI briefs?
It Does What Now? describes its briefs as "short neutral" entries that link to their original sources. Rather than relying on a single outlet, it aggregates the news of the AI era into a dated timeline and points each item back to the primary reporting or announcement it came from.
What that looks like in practice
- Each entry is a brief with a date and a link to its original source, so you can trace a claim rather than take the summary on faith.
- The timeline runs from January 2020 onward and is organised by year, with tracks such as Models & capabilities, Safety & alignment, Government & policy, Courts & copyright, and Money & business.
- Alongside the news entries, it carries a separate commentary section that names the writers and publications being quoted — for example Don't Worry About the Vase, Transformer, and Epoch AI.
Why the sourcing matters for how you use it
If you are writing a briefing, teaching a course, or trying to reconstruct when a particular argument or policy shift happened, a dated timeline with per-item links is more useful than a single explainer, because you can jump from the summary to the underlying document. The trade-off is that a brief is only as good as the link behind it: the site tells you what happened and where it came from, but the depth and framing of each item vary with the source it points to.
A practical next step
Pick one entry that affects your work — a policy order, a model release, a court ruling — and follow its link to the primary source before citing the timeline itself. For a second opinion on the same event, check whether the brief points to an official announcement or only to press coverage, and compare it against a general news outlet such as Reuters.
Is there a newsletter or subscription for updates from It Does What Now?
Yes. The site links to a newsletter on Substack, and that is the main way to get updates rather than checking the timeline page manually.
What the newsletter is for
It Does What Now? is a running, sourced timeline of AI news from January 2020 onward, organised into tracks such as models and capabilities, safety and alignment, government and policy, money and business, and compute and infrastructure. New entries are added as events happen. The newsletter is the natural companion: it pushes new entries to you instead of you scrolling the timeline.
How to decide
- Subscribe if you want a chronological, source-linked record of AI developments and would rather receive updates than revisit the site.
- Skip it if you only need occasional reference — the site itself is browsable by year and by track, and the "story so far" view covers the whole era in one sitting.
- Check the pricing page before committing, since the linked page is the subscription sign-up and terms may change.
Next step
Open It Does What Now? and use the newsletter link near the footer to see current options, then compare that against simply bookmarking the timeline if you prefer reading on your own schedule.
Cloudflare
User reviews (0)