DATA TRAINING DAZE : ICEBERG AHEAD! A.I. ESSAY

PROFESSIONALS are training A.I. to do their jobs. Start-ups are paying to have bots acquire increasingly rarefied skills.

EVERY DAY, MERCOR, a start up that sells training data to artificial intelligence companies, pays 30,000 contractors more than $4 million to help make their job, and those of their colleagues, obsolete.

It's gig work, but for professionals with rarefied skills. One recent Mercor posting offered $225 an hour for a voice actor able to maintain a customer service persona in fluent Hebrew.

Another sought a PhD physicist with a specialisation in general relativity, astrophysics or cosmology. A third listing wanted a physician with more than three years of experience in the Rwandan primary care medical system.

Mercor and a handful of similar startups are the primary middlemen in a supply chain of '' human data '' that may power the next generation of A.I.

As OpenAI, Anthropic and other major ventures compete to become the industry's dominant platform, the market for premium data that has been vetted by experts is exploding.

No longer do the A.I. companies need armies of low-keyed workers often overseas, to do rote tasks like tag images of cars or transcribe audio. They need mathematicians to annotate proofs, lawyers to mark up briefs and professors to grade essays. 

That's what Mercor and its rivals supply. To use the parlance of the industry, data labeling has moved up the '' value chain, '' and the start ups that offer this service have become some of the fastest growing in Silicon Valley.

Mercor, which was founded in 2023, announced a funding round last October at a $10 billion valuation, last week, Bloomberg reported that the company is now talking to investors about a deal at twice that level.

Last year, Meta invested more than $14 billion in another data training venture, Scale AI, in part to hire its chief executive.

HANDSHAKE, a recruiting start-up that pivoted to data training venture, only last year, says its annualized revenue rate crossed $1 billion in April, up from $550 million at the start of the year.

These data-traing start-ups are exploiting a market opportunity - selling to well capitalized labs a product for which there is now near-unlimited demand.

But it's a delicate moment. The training start-ups need ChatGPT, Claude, and other A.I. models released by their clients to keep improving, to demonstarte that they are adding value.

They also need the models to remain imperfect so that these clients keep coming back for more data.

This Master '' ICEBERG AHEAD '' Essay Publishing, continues. !WOW! thanks Lora Kelley.