How are we faring in AI boom?

It’s an intuitive assumption, given a vibrant startup scene typically attracts younger employees, who prefer the city over the suburbs.

“There’s also a lot of free venue space for events, because of San Francisco’s 40% vacancy rate for commercial. As a result, there’s numerous AI events that are being hosted at very low cost,” said Blitzscaling Ventures General Partner Jeremiah Owyang.

But it’s also reflective of the difference in the new crop of AI firms compared to traditional technology companies, as the former simply need fewer people to operate. The term “AI” is broad and is often used when referring to anything from image generator apps to complex foundational models — think OpenAI’s GPT-4 or Meta’s LLaMA 3. And the companies receiving the most funding are focused on developing the latter, rather than just using them for application purposes, said Owyang. That means their funds are geared not toward hiring new workers, but rather on compute power, which requires expensive access to high-performing, energy-intensive data centers.

Industry leader Anthropic has raised several billion dollars over the last two years, and it would have likely added a lot more employees to its payroll if it were a traditional tech company. The same is also true for Perplexity, which has less than 100 employees, as well as OpenAI, estimated to have around 700.

“As far as we know, [OpenAI] has under a thousand people, and that’s the leader. The leading AI company in the world has less than a thousand people, and they’re not a new startup. They’ve been around,” Owyang said. “This is an important thing for society to wrestle with, that really efficient companies with AI just don’t need a lot of human workers.”

Div Garg, co-founder of MultiOn, a Peninsula-based startup, said his eight-person team spends a large chunk of their money on researching and developing models, which includes computing costs.

“Traditionally, hiring and people costs are the biggest expense in running a startup but, right now, what’s starting to happen is that compute power is becoming the biggest expense, more than the people expense,” he said.

Real estate, labor effects

The effects on the local labor and commercial real estate markets are starting to bubble up too. Office vacancy rates are still high, and the total number of jobs have seen a net decrease in San Francisco and San Mateo counties between April 2023 and April 2024 — especially for professional and technical roles — largely attributable to widespread downsizing and less venture funding. And some jobs will likely see an especially painful recovery, if they’re not permanently altered.

While traditional AI enabled automation to replace relatively low-paid jobs, with advances in AI, particularly generative AI, jobs that are currently performed by high-paid knowledge workers will start to be replaced, TDK Ventures Investment Director Henry Huang said in an email.

The share of AI jobs as a percentage of all technology jobs in San Mateo County jumped from 22% in April 2023 to 39% in April 2024, with the total number of AI roles also doubling. San Francisco’s ratio of AI-to-tech jobs also increased from last year to this year, according to data from the University of Maryland.

Matt Huang, co-founder and chief product officer of Cofactory, said he now expects certain employees, like marketers for instance, to execute what usually would comprise multiple roles.

“I’d rather have one person who can do headlines, performance marketing campaigns, also do landing pages and set up campaigns across different channels,” Huang said. “I expect one person to do what might have been four separate jobs back in the past.”

While they might not contribute heavily to short-term job creation, both Huang and Garg said their products are important tools that free up time in people’s personal and professional lives. MultiOn is an AI agent that can complete mundane tasks with a simple request, such as booking a flight or making a restaurant reservation.

“We don’t want our agents to be replacing humans, but we want them to collaborate and do a lot of the manual jobs that we don’t want to do,” Garg said. “It’s almost like a copilot.”

Government oversight

Just like any other new technology, however, the quick pace of AI’s evolution leaves some individuals less optimistic, with elected leaders scrambling to figure out how best to regulate it and stem potential job losses. Earlier this year, the San Mateo County Board of Supervisors adopted a resolution stating AI won’t replace any county employee’s position until they leave for an unrelated reason. And state and national leaders have attempted, with mixed success, to legislate its effects, a Herculean lift given its seemingly ubiquitous presence and nebulous definition.

“I would say regulation is not keeping pace,” said Jill Finlayson, managing director of the CITRIS Innovation Hub in Berkeley. “For example, we have rules in hiring practices that say you can’t be discriminatory … but when you bring in an AI tool, and individuals just get an individual response, they don’t know if there’s a pattern of bias.”

The Pleasanton-based software vendor Workday was recently the subject of such a controversy, with a lawsuit alleging its AI tool discriminated on the basis of race, age and disability as part of its screening process.

“At the end of the day, the hiring party is responsible for any bias, but if they’re using a third-party tool, and that third-party tool guarantees that they’re not going to discriminate, and then it does, who’s accountable?” Finlayson said.

But San Mateo County’s relatively small share of the AI funding in the Bay Area could actually be advantageous in that regard, allowing for a more measured regulatory approach.

Garg said MultiOn’s proximity to Stanford University has helped with hiring, but it also poses challenges because many involved in the AI space want to be in San Francisco instead. But that won’t be the case forever, particularly around Palo Alto — right behind San Francisco in terms of AI funding — and Menlo Park, Owyang said. AI startups are popping up around Springline, a new Menlo Park development, and the CEO of AI firm Replit recently announced it’s moving from San Francisco to Foster City, citing his desire to base the company in a more “livable” city.

“As these AI companies evolve and find market traction, they want to set up a proper headquarters. In some cases they want to do it in the area where the quality of life is better, and that’s the Peninsula,” Owyang said. “So we will see an increase of the more mature AI companies heading our way.”

Peninsula impact

And while AI could drastically alter some knowledge workers, such as marketers or human resource employees, the Peninsula’s strong biotechnology and pharmaceutical sector may not feel such drastic effects, despite robust AI investment. Kevin Parker, co-founder and CEO of South San Francisco-based Cartography Biosciences, feels the industry holds a little more of a buffer against potential job automation than the tech industry, as any gains made from AI — whether it’s related to drug discovery or accelerating clinical trials — must still be paired with an intricate network of scientists and researchers involved in a long and winding drug approval process.

“You still need somebody upstream to generate data and downstream to test drugs … It’s adding more emphasis on computational teams, but I think it’s net creating jobs, not taking them away,” Parker said, adding that its impacts will likely be more incremental than sudden. “It’s sometimes portrayed too much as a binary, but it doesn’t have to be an all-or-nothing type of thing.”

alyse@smdailyjournal.com

(650) 344-5200 ext. 102