This article originally appeared on Business Insider.
Recruiting AI talent can be difficult for some companies.
Aravind Srinivas, founder and CEO of Perplexity, an AI-powered question-and-answer engine, described an interaction he had with a job candidate that illustrated how difficult it is to hire people with generative AI skills.
“I tried to hire a senior researcher at Meta, and you know what they said? 'When you get 10,000 H100 GPUs, come back to me,'” Srinivas says. he said on a recent episode of the business advice podcast Invest Like the Best. ”
H100 GPU refers to Nvidia's highly coveted graphics processing unit that tech giants like Meta, OpenAI, and Google use in their data centers to power and train their AI chatbots. Masu.
“It would cost billions of dollars to make it happen, and it would take five to 10 years to get it from Nvidia,” Srinivas said.
Perplexity, which uses GPT-4 to power its Q&A engine, says limited funding and a lack of chips are making it difficult to find the talent needed to create language models at scale. said.
Srinivas said it's difficult to get employees to leave a company that “has a great experiment stack and an existing model to bootstrap from.”
“We have to provide such great incentives and immediate availability of computing. And we're not talking about small computing clusters here,” he said. .
The CEO added that even if small businesses like Perplexity were able to get Nvidia chips, they would continue to fall behind because AI is developing so quickly.
Srinivas said AI talent at big tech companies “will already be building next-generation models.”
“They're like, 'Look, the world has changed, I'm already in the next generation,'” he added. “'I'll come back when I've finished training the next version of the model. This time I'll have H100. Come back when I get 20,000.'
Srinivas and Mehta did not immediately respond to Business Insider's requests for comment before publication.
Since OpenAI launched ChatGPT in November 2022, interest in AI skills such as machine learning and data engineering has rapidly increased. Companies like Amazon, Netflix, and Meta are offering salaries as high as $900,000 per year to attract generative AI talent. Technology companies in education, healthcare, and law are looking for people who know how to use AI.
Srinivas believes that workers need skills beyond the ability to create AI models that produce desired outputs.
“We need to train them after the fact and address the long tail of issues that arise as we deliver the product,” the CEO said.
Srinivas said post-training expertise, such as how to reduce factual inaccuracies in chatbots, is a key skill that employees across a wide range of digital industries can quickly learn.
Leveraging that skill set will help AI companies like Perplexity stand out in a field dominated by Big Tech, he said.
“You have a huge advantage of creating a lot of value,” he said of the post-training skills. “And that's what we're focused on.”
