One AI training incumbent missed the mark on key aspects of the business, according to a rival CEO.
On an episode of the “20VC” podcast released on Monday, Mercor CEO Brendan Foody said that Scale AI’s former CEO, Alexandr Wang, was “phenomenal” at aspects like distribution and sales, but the startup sidetracked in other areas.
“In some ways, Scale lost the focus on product, on scaling quality,” Foody said. “And that was one of the largest challenges of the business.”
In a statement to Business Insider, Joe Osborne, a Scale AI spokesperson, said: “There must be something Brendan admires about Scale if he keeps mentioning us to get publicity. Our data quality metrics are at record highs and we remain focused on leading the market.”
Mercor did not immediately respond to request for comment.
Data annotation startups like Scale AI, Mercor, and Surge AI pay hundreds of thousands of part-time contractors around the world to filter, rank, and train AI responses for the world’s largest AI companies.
The Mercor CEO, who cofounded the company in 2023, said that the most important aspect of the business was quality and “having phenomenal people that you treat incredibly well.”
He added that Mercor’s average pay rate is $95 an hour, while he said rivals like Scale and Surge pay about $30 an hour. Pay rates for AI training tasks vary based on the location and educational qualification of the contractor, as well as the complexity of the task.
Scale pays STEM experts about $30 to $50 per hour and generalists about $15 to $30 an hour, per its website. Mercor’s website says it pays STEM experts about $90 to $110 an hour, and generalists about $45 an hour.
In June, Mercor’s head of product, Osvald Nitski said the company is focused on recruiting elite-level annotators, like International Math Olympiad medalists, Rhodes Scholars, and Ph.D. students.
Foody’s remarks come after big shake-ups at Scale AI, which received a $14.3 billion investment from Meta in June. Following the investment, the startup scrambled to convince investors and other Big Tech clients that their data remains secure despite its close entanglement with Meta.
The same month, Business Insider reported serious security holes in the company’s practices. Business Insider found that Scale AI routinely used public Google Docs to track work for high-profile customers like Google, Meta, and xAI. That practice left AI training documents labeled “confidential” accessible to anyone with the link.
At the time, Scale AI said it takes data security seriously and locked down public access to the Google Docs.
In July, Scale AI said it laid off approximately 200 full-time employees, or about 14% of its 1,400-person workforce, and 500 contractors.
“While that felt like the right decision at the time, it’s clear this approach created inefficiencies and redundancies,” Scale’s interim CEO Jason Droege wrote in an internal email about the layoffs, which was viewed by Business Insider. “We created too many layers, excessive bureaucracy, and unhelpful confusion about the team’s mission.”
Read the full article here
