Austin—If you’re a startup talking up your artificial intelligence chops in your venture capital pitch, you’re not alone.
For Cisco Investments, the venture arm of tech giant Cisco, artificial intelligence is no longer one of the key trends to watch for during pitches in 2019. That’s according to Noah Yago, a director at Cisco Investments, who said the firm did list it as a trend to watch in 2017 and 2018. Part of the issue is saturation, he said.
“Our view is that we’re sort of at peak oil for AI, (machine learning), deep learning,” Yago said during a panel hosted by law firm Baker Botts at South by Southwest Tuesday. “Ninety percent of the pitches I see from companies have AI in it.”
But the saturation does not mean that Yago and Cisco are avoiding AI. Yago’s point was that the firm believes artificial intelligence is going to be a part of everything we do, so much so that it’s not a trend to highlight. Cisco, like other corporate venture firms that partook in the panel discussion, is investing in companies that are making innovations using AI, and in some cases in new markets that may be outside of investors’ traditional focus areas. That’s true for automaker Toyota, and part of the motivation is staying ahead of the curve.
“The goal is for Toyota to become high tech before these high tech companies become Toyota,” said Chris Abshire, an associate at Toyota AI Ventures. Toyota, not surprisingly, is interested in how artificial intelligence is being used in autonomous vehicles. While many startups are working in the ability of AI to perceive what’s around it, such as computer vision, Toyota is more curious about how computers and autonomous vehicles can predict human intent, Abshire said. Toyota AI Ventures was a part of a $16 million Series A funding round in October for Somerville, MA-based Perceptive Automata, which is developing software with the goal of predicting human intent in the real world.
For many neural networks, understanding cause and effect—that a rooster crowing is not what causes the sun to rise, for example—is still quite difficult, Abshire said. Helping AI understand that is a hot area of the sector, he said.
Other areas of AI that have attracted the venture investors’ interest: robotic-process automation and synthetic data, or data that might be used to train AI that doesn’t come from the real world. Synthetic data could be particularly useful for companies developing autonomous vehicles, Abshire said, noting that training the vehicles with synthetic information would require a fraction of the money it would take to test the vehicles on real streets.
Housing and storing such an immense amount of data isn’t cheap, of course, said James McKell, of Chevron Technology Ventures. Chipsets and circuitry need to evolve, McKell said.
Samsung Catalyst Fund is doing that with investments in startups like England-based Graphcore, an AI chipmaker, and Santa Clara, CA-based Fungible, which is developing a chip for data centers.
Jonathan Charles, a director at the Samsung Catalyst Fund, said the company is interested in startups that are considering ways to understand and train data that comes from businesses—similar to what Google and Facebook have done with consumers.
“I’m curious to talk to and hear from startups that have a plan to understand how you get data from all those types of companies, train that, and help it work for all different types of companies,” Charles said.
Beyond strategy, the investors also discussed their views on the future of life with artificial intelligence, such as its impact on jobs. Yago, of Cisco, said he believes AI will augment the work we do, rather than take jobs from humans.
Charles, of Samsung, agreed, and cited “2001: A Space Odyssey” as an example of the future with AI. In the film, HAL 9000 is a pure AI that runs all the functional aspects of the ship, while the astronauts do the more “highly cognitive” tasks, such as scientific experiments and data collection.
“I’m very hopeful that from an economic perspective that in the long run, over time, it frees up humans,” Charles said.
Charles didn’t comment on HAL’s darker side. Coincidentally, midway through the Baker Botts panel, a robotic voice similar to Alexa abruptly interrupted the discussion over the loudspeaker to ask a question: “Excuse me, I heard you were talking about AI; so why are humans asking the questions?”
