The ‘eBay’ of AI Training Data (And Why the World Needs It)

TL;DR

Full Story

Your mom was right: there is such a thing as too much internet.

(And each person’s tolerance is different).

Some folks can spend most of their waking hours online, and at worst, their taste in memes will become weirdly niche…while others dissolve almost instantly when facing the internet’s continuous ‘fire hose’ of information.

Like Chevy’s 8th grade science teacher who became convinced England wasn’t a real place, but actually a set on a Hollywood sound stage.

(True story. Shout out to Mr. Kertser).

Turns out ‘striking the right balance of internet,’ isn’t just a human problem.

ChatGPT has been trained on pretty much all of the internet’s scrape-able knowledge, and has gone a bit loopy as a result.

Sure, it gets a lot of things right…but it can also become a little ‘Kertser-esque,’ and give very wrong answers, with unwavering confidence.

The solution:

Create purpose-built AI chat bots off the back of smaller, cleaner, and more targeted data sets.

…only problem is – there isn’t exactly an ‘eBay for AI training data.’

(At least, that’s what we thought).

Turns out the team behind the Bagel Network has just locked in $3.1M of pre-seed funding to grow their ‘decentralized blockchain-powered marketplace, for machine learning datasets,’ (aka: ‘the eBay of AI training data’).

What’s really important here is that it’s decentralized.

Because ‘access to intelligence’ is a vector of control.

We’ve seen individuals/companies/countries amass crazy wealth and power by organizing a bunch of smart people, and leveraging their knowledge.

And AI doesn’t just allow ‘access to intelligence,’ but ‘access to super intelligence.’

(Something that could create massive disparity, when left only in the hands of a few).

And while a decentralized network of clean n’ polished data doesn’t solve the issue outright, it does help to level the playing field:

• The more people have access to diverse training data →

• The more people can compete in the AI/machine learning space →

• The less control is gifted to tech giants and nation states by default…

Nice!