diff --git a/content/ai-metascience/index.md b/content/ai-metascience/index.md index 1259328306e..828881e0bbb 100644 --- a/content/ai-metascience/index.md +++ b/content/ai-metascience/index.md @@ -252,9 +252,35 @@ image: } .am-prog-talks { list-style: none; margin: 0; padding: 0; } .am-prog-talks li { + display: flex; + align-items: center; + gap: 0.5rem; padding: 0; border-top: 1px solid var(--am-rule); } +.am-prog-watch { + display: flex; + flex-wrap: wrap; + justify-content: flex-end; + gap: 0.3rem; + flex: 0 0 auto; + max-width: 11rem; +} +.am-prog-watch a, +.am-prog-playlist a { + padding: 0.15rem 0.55rem; + border: 1px solid var(--am-accent); + border-radius: 999px; + color: var(--am-deep); + font-size: 0.76rem; + white-space: nowrap; +} +.am-prog-watch a::before, +.am-prog-playlist a::before { content: "\25B6\00A0"; font-size: 0.65rem; } +.am-prog-watch a:hover, +.am-prog-playlist a:hover { background: var(--am-tint); text-decoration: none; } +.am-prog-playlist { margin: -0.4rem 0 0.9rem; } +.am-prog-playlist a { font-size: 0.86rem; } .am-prog-talks li:first-child { border-top: 0; } .am-prog-item { display: flex; @@ -345,6 +371,8 @@ image: .am-prog-item { flex-wrap: wrap; gap: 0.3rem; } .am-prog-time { flex: 0 0 100%; } .am-prog-text { flex: 1 1 100%; } + .am-prog-talks li { flex-wrap: wrap; } + .am-prog-watch { flex: 1 1 100%; max-width: none; justify-content: flex-start; padding: 0 0.4rem 0.5rem; } } .am-prog-dialog { @@ -485,43 +513,44 @@ Here are the sessions and the talks in each, in the order they run. **Click any

Monday 28 SeptemberAI in the research pipeline

+

All Monday recordings on YouTube

Morning

  • 09:30Welcome and framingThe organisers

09:40When AI writes the literature

The problem as it already shows up in the correction record, the pressure producing it, and two routes to a better literature.

    -
    • -
    • -
    • -
    • +
  • Watch
    • +
  • Watch
    • +
  • Watch
    • +
  • Watch

11:10Evidence synthesis: capability, comparison, assurance

From what frontier models can extract, through a head-to-head test of the tools built on them, to the safeguards that would make an AI-assisted review checkable.

    -
    • -
    • -
    • +
  • Watch
    • +
  • Watch
    • +
  • Watch

Afternoon

13:15Evaluating science, and its evaluators

What it takes to make a paper checkable in the first place, what human review actually does, what a computational novelty signal might add, and a tool that checks a paper against what its authors said they would do.

    -
    • -
    • -
    • -
    • +
  • Watch
    • +
  • Watch
    • +
  • Watch
    • +
  • Watch

15:40Conditions for legitimate adoption

What researchers do with these tools, what open infrastructure and open data would require, who gains or loses when the tools arrive, the rules of use that follow, and what it would take to make research outputs readable by machines in the first place.

    • -
    • -
    • -
    • +
  • Watch
    • +
  • Watch
    • +
  • Watch

17:10Closing provocation

@@ -532,20 +561,21 @@ Here are the sessions and the talks in each, in the order they run. **Click any

Tuesday 29 SeptemberReplications and reproductions

+

All Tuesday recordings on YouTube

Morning

09:30Symposium: Can AI Help Us Understand How Findings Move from Animal Studies to Human Studies?

Four talks, mostly by early-career researchers, on tracing the biomedical evidence trail across publications, trial registries and regulatory documents. Chaired by Benjamin Ineichen.

    -
    • +
  • KoberChopardDonevaIneichen

11:15Reproducing the record

Replication and reproduction as work actually done, with agents in the loop and with the paperwork that makes a result reusable, followed out to whether anyone downstream notices.

    -
    • -
    • -
    • +
  • Watch
    • +
  • Watch
    • +
  • Watch

Afternoon

@@ -558,23 +588,24 @@ Here are the sessions and the talks in each, in the order they run. **Click any

14:40Predicting replicability

Three talks on forecasting which findings will hold, and what such forecasts are for.

    -
    • -
    • -
    • +
  • Watch
    • +
  • Watch
    • +
  • Watch

16:05Machines that build and test the replication record

First the benchmarks that measure what these agents can actually do, then the two databases being built on that basis with humans kept in the loop, and last the question the day has been circling: can an agent run the checks at human quality?

    -
    • -
    • -
    • -
    • -
    • +
  • Watch
    • +
  • Watch
    • +
  • Watch
    • +
  • Watch
    • +
  • Watch

Wednesday 30 SeptemberWorkshops, and analyst variability

+

All Wednesday recordings on YouTube

Morning

    @@ -583,19 +614,19 @@ Here are the sessions and the talks in each, in the order they run. **Click any

    Afternoon

      -
      • +
    • Watch

    15:00Does the finding survive another analyst?

    The other way a result fails to hold. The study is not repeated badly; a defensible alternative analysis simply lands somewhere else. And the alternatives increasingly consult the same model.

      • -
      • +
    • Watch
      -
      • +
    • Watch
@@ -712,7 +743,7 @@ Everything runs on Google Meet. We will email joining details to everyone who ha Live captions are on in every session. If anything else would help you take part, tell us on the registration form, and we will do what we can. -Where speakers opt in, their talks are published afterwards on the [FORRT YouTube channel](https://youtube.com/@forrtproject9995). +Recordings of the talks whose speakers opted in are on the [FORRT YouTube channel](https://youtube.com/@forrtproject9995), with one playlist per day: [Monday](https://www.youtube.com/playlist?list=PLS3VdaYKcrew), [Tuesday](https://www.youtube.com/playlist?list=PLSItTTcc91ac) and [Wednesday](https://www.youtube.com/playlist?list=PLZYxy57rykQU). The **Watch** links in the programme above open each talk. ## Code of conduct