Dead startups become gold for AI training

Failed companies' data is now a commodity. AI labs are buying codebase and workspace records from defunct startups, but the practice raises legal and ethical concerns.

10 cards · 2m · tap to begin

From · · 2m

Dead startups become gold for AI training

Failed companies' data is now a commodity. AI labs are buying codebase and workspace records from defunct startups, but the practice raises legal and ethical concerns.

In brief

Failed companies' data is now a commodity. AI labs are buying codebase and workspace records from defunct startups, but the practice raises legal and ethical concerns. AI labs are purchasing data from defunct startups to train their models as public datasets run dry, creating a new market but raising legal and ethical…

Crypto founder turns shutdown into cash

When Keelan Miskell wound down his crypto startup, he discovered its data was valuable. He sold the company's codebase and workspace data for tens of thousands of dollars.

A market for dead startup data

Companies that help shut down startups now sell their data to AI labs. This growing market emerged as training labs exhausted public and synthetic datasets.

AI labs seek real-world records

AI labs seek real-world records

Buyers want real-world records to train AI agents. The demand reflects how training labs have hit the limits of publicly available and artificially generated data.

Legal and ethical questions loom

The sales of startup data raise legal and ethical concerns. The article does not specify what those concerns are, only that they exist.

Animals have much to tell us

Humans are far from the only species with a gift for gab. Our furry, feathered, and finned friends could have a lot to tell us about their inner lives.

Decoding animal communication

Decoding animal communication

Decoding how animals communicate could reveal their inner lives and how they see the world. A growing arsenal of devices and AI advances have made studying their howls, squeaks, and clicks easier.

Complexity remains in translation

Making sense of animal communication remains complex despite technological advances. The article notes progress but does not claim the problem is solved.

Moon bases face hostile conditions

The US and China are racing to build lunar bases. NASA hopes to establish a small lunar city by the end of the century despite the moon's hostile environment.

Survival on the moon demands innovation

A lunar base must protect inhabitants from asteroids, cosmic radiation, razor-sharp dust, and extreme temperature swings. It must also be somewhat self-sustaining.

Technology and politics collide

Building a lunar base will require autonomous robots, nuclear power, and ways to use the moon's resources. But the chief obstacle may be one of political will.

“the chief obstacle may be one of political will”

The takeaway

AI labs are purchasing data from defunct startups to train their models as public datasets run dry, creating a new market but raising legal and ethical questions.

Read the original on MIT Technology Review

React

Sign in to react and comment.

Comments (0)

Life is short. Keep it sweet. Respect others' opinions and be kind!

    Recommended next

    More decks on ai training and related topics.