Here are two scenarios for where artificial intelligence might be headed:
#1: Companies selling technology to support more advanced AI — and those applying it quickly — take over the economy, sending many older and less nimble companies into a death spiral.
#2: Those companies dominate the economy for a bit, and then AI becomes so powerful that it subjugates or exterminates the human race.
We should be talking more seriously about the latter. The dangers of AI are one reason that thousands of leaders, including researchers at MIT, Harvard, and Northeastern, signed an open letter last month proposing a six-month pause on developing systems more advanced than OpenAI’s GPT-4. I’ll explore why serious people are worried about that scenario in a future piece.
But if you’re not (yet) fretting that ― after coping with COVID ― we now need to keep HAL 9000 from trying to kill us, let’s talk business. Boston has been a center of AI research since the 1950s, and in the 1980s there was briefly a part of Kendall Square dubbed “AI Alley,” home to hardware and software companies like Thinking Machines and Symbolics.
At the moment, though, much of the activity and funding around advanced AI is in the Bay Area. After attending a San Francisco conference sponsored by the venture capital firm Sequoia last month, Brian Halligan, cofounder and chairperson of Boston software company HubSpot, wrote on Twitter: “I learned a lot by overhearing conversations at crosswalks. Boston: Celtics, politics, weather, exercise, food, etc. SF: AI, AI, AI, AI, AI, etc.”
When I caught up with Halligan, he didn’t want to talk much on the record, for fear of sounding too negative. But he noted that San Francisco’s upside mentality about tech — honing in on trends, and pouring money onto them like gasoline on a campfire — was at play. Meantime, he said, Boston was not doing enough to spawn and support so-called “generative AI” companies, which can produce new types of text, music, and images in response to typed commands. “We have a lot of smart people,” Halligan said, and an opportunity to create a next generation of valuable companies, perhaps even bigger than HubSpot.
On a list of roughly 90 “unicorn” companies working on AI, compiled by the research firm CB Insights, just three were based in Massachusetts, according to Drew Volpe, a Cambridge venture capitalist who has made several investments in the sector. (A unicorn is a privately held company valued at more than $1 billion, according to investors’ estimates.)
“There’s a ton of interest in the VC community, but most are still figuring out where the right bets will be,” said Volpe, managing partner at First Star Ventures. Some are worried about whether these new AI companies will compete with software firms, and even earlier generations of AI companies, that they’ve already put money into. Some of the new companies may simply prove to be “thin wrappers” around the powerful language-processing models that OpenAI has built, and not create enough value on their own, Volpe said. “My own view is that there will be huge winners who aren’t ‘AI’ companies, but leverage these new interfaces in the same way the rise of smartphones allowed Uber to create ride sharing,” he said. “Boston is well positioned to create many of these, given our strengths in industries like health care, robotics, and climate.”
Greg Raiz also sees similarities between AI and the boom in mobile. He built and sold a company, Raizlabs, that helped develop mobile apps for companies that suddenly realized how much time people were spending gazing at their phones. Following the sale, he went on to run the Techstars Boston accelerator program, which provides funding and mentorship to fledgling ventures.
Now, he’s planning to launch an accelerator, XLR8, that will support companies focused on opportunities around the latest AI technology. “Generative AI is moving at light speed,” Raiz said. “Young companies can pivot and move and adapt much faster than large incumbents.” His plan is to identify a first batch of between five and 10 promising concepts, put $25,000 of his own money into each, and get the accelerator’s mentors and advisers to kick in another $75,000 per company. That funding will buy between 2 and 4 percent of a startup’s equity, Raiz said. While much of the programming to support the companies’ growth will be delivered online, Raiz is planning to hold a kick-off event in Boston this summer.
On Thursday, events impresario John Werner is hosting a one-day conclave called Imagination in Action: Forging the Future of Business with AI. It will feature speakers from Fidelity, Prudential, and Salesforce, along with OpenAI chief executive Sam Altman and Tesla cofounder Marc Tarpenning. Next week, Northeastern University hosts a gathering on the “creative potentials and ethical responsibilities” of generative AI.
In May, Judah Phillips, founder of the AI startup Squark, convenes the first-ever Boston Generative AI Meetup. Entrepreneurs Rob May and Abhi Yadav are building the AI Innovators Community, and hosting occasional invite-only dinners. A new program at MIT’s Computer Science and Artificial Intelligence Laboratory, the T-Minus Fellowship, is offering early funding to students, faculty, and staff building AI-related startups.
If you’re betting on Scenario #1, and hoping we can stave off Scenario #2, there is a massive amount of economic value that AI firms are about to create. In the same way Boston is already leading in the next generations of robotics, we need to be focused on AI competitiveness as well.
“We need to get in gear,” said. Raiz. ”There’s a lot of opportunities for Boston to get the community to rally around this stuff. The research and the talent here is incredible. We just have to apply it.”