Handshake's CEO says the AI training world is evolving from generalists to STEM experts getting paid over $125 an hour
Briefly

Handshake has transitioned from a recruiting platform to an AI training service. CEO Garrett Lord emphasized the need for specialized math and science experts in data annotation instead of generalists. He noted that as AI models advance and incorporate vast information, generalist roles are becoming obsolete. Contractors on Handshake are making substantial earnings compared to generalists on other platforms. The CEO's comments follow significant developments in the AI sector, including Meta's investment in Scale AI, which has influenced the demand for specialized knowledge.
Garrett Lord, CEO of Handshake, stated that the data annotation industry is shifting from generalists to needing highly specialized math and science experts. 'They've gotten good enough where like generalists are no longer needed.' This indicates a significant evolution in AI training demands, requiring advanced subject knowledge in areas such as accounting and law, in addition to STEM fields like physics, math, and chemistry.
Contracts for domain experts on Handshake pay between $100 to $125 per hour, reflecting the increasing need for specialized knowledge in AI training projects. In contrast, generalists earn much less, between a few dollars to about $40 per hour, depending on the task and location.
Meta's substantial $14.3 billion investment in Scale AI has prompted changes in the competitive landscape of AI services, notably prompting Google to halt several projects with Scale. Additionally, other companies like OpenAI and Musk's xAI have paused work with Scale, highlighting the volatility in AI provider partnerships.
The CEO recognizes that as AI models improve and become more integrated with vast data sources from the internet, the requirement for extensive domain knowledge increases. This trend signifies a niche specialization in the tech job market, reflecting the evolving nature of AI applications.
Read at Business Insider
[
|
]