“We take a recent AI research breakthrough … we give AI systems a metric … and we see if it can replicate that human innovation; the tripwire sounds if it does”

The TWIML AI Podcast (formerly This Week in Machine Learning & Artificial Intelligence) · From Math Olympiads to Navier-Stokes: How Fast Is AI Progressing? with Greg Burnham

Play from 17:32
17:32

Audio streams from the publisher’s own feed. Verbatim does not host or copy it.

What was described

Built a 'tripwire' benchmark that gives AI systems a pre-innovation metric to improve, without internet access, to test if they can replicate recent human AI research breakthroughs

Said nearby on the same subject

“Number two, lawyers need to educate members of the firm about the difference between traditional legal research using Westlaw and the use of generative artificial intelligence”
♫
Said on Bloomberg Law
How they work Share this moment
“We tested, OK, let's have humans provide inputs upfront on some context elements that the AI cannot have access to. And then let's let the AI do the job that we got the best results”
VB
Vincien Borchene
The So What from BCG ·
How they work Share this moment
“No chat bot or any type of artificial intelligence can be advertised as being able to be used for mental health or therapy without a provider involved”
♫
Said on A Beginner's Guide to AI
Why it worked Share this moment
“If you have a physician that prefers to use AI, Open Evidence is a medical AI site… it will outline the recent research I have found in a very clear way.”
JS
Jessica Slice
Full-Tilt Parenting: Strategies, Insights, and Connection for Parents Raising Neurodivergent Children ·

Moments like this, every Monday

The strongest new moments on your topics, in the speakers' own words. Free for three topics.

No card. No password. One-click unsubscribe.

Track a competitor?

Pro watches up to 25 names or phrases and emails you the morning after they are said.

See Pro, $29 a month

Verbatim quotes a short extract and links to the episode; the episode belongs to its publisher. Remove a show.