There are many burgeoning opportunities in artificial intelligence, but venture capital investors during an SXSW panel agreed they all hinge on targeted applications and access to data.
Since most of the data in the world is proprietary or regulated, it can be difficult for small companies and startups to access enough data to properly test and demonstrate their AI products, so access to a large data set is very helpful for VC interest.
Kristina Serafim, investment director for Verizon Communications Inc.'s venture fund, said even in partnership arrangements with smaller companies, data access can be challenging, and for this reason Verizon Ventures usually partners with and invests in AI firms with the intent to eventually acquire.
"Our advantage is absorbing some of the companies and integrating [them] into our own data sets," she said.
In many industries ripe for AI efficiencies, access to data is complicated by government regulations. This is particularly true in the healthcare and financial services spaces. There is a "huge opportunity" for AI applications in those fields if the data access problem can be solved, Aaron Jacobson, partner at tech and healthcare venture capital firm New Enterprise Associates said during the panel.
His firm and the other panelists agreed that focusing on a targeted, solvable problem is also key.
General AI, in which a machine makes a wide range of generalized decisions in a variety of environments, is likely decades away, Jacobson said.
"General purpose AI is overhyped. That's not what VCs are looking for," he said.
Rather, firms are focused on specific, narrow applications in any industry, but largely focused on white-collar jobs. Where automation revolutionized blue-collar manufacturing in the 20th century, automation in the 21st century will revolutionize the white-collar workforce, the panelists agreed. For an AI startup looking for funding, a targeted application that addresses a specific enterprise problem is key.
For example, Jacobson was critical of autonomous vehicle AI startups. While there are some good targeted ideas in that industry for nascent applications, the problems with self-driving cars are generally vast and complicated. Commercial availability of self-driving vehicles is probably at least a decade out, he said.
AI solutions should not necessarily be targeted at replacing the human workforce, he added. His firm is looking more for AI that optimizes the human workforce. When ATMs were invented, people speculated that they would replace human bank tellers, but they only freed up human tellers to do more complicated tasks and customer service, he said.
"The investments we get excited about are making existing humans more effective," he said.
In general, the AI opportunity is massive, panelists agreed. Jacobson compared it to cloud technology 10 years ago. David Blumberg, founder of Blumberg Capital Management LLC, meanwhile, compared it to the social media opportunity 15 years ago. There is a low penetration of the technology now, with companies spending about 10% of their IT budget on AI solutions, Jacobson said. He predicts that to grow to as much as 40% by the time the AI matures.
Blumberg said about 90% of the data that is collected across governments and businesses is not even used or applied. He further said the amount of data out there is rapidly swelling as internet of things offerings and other connected technologies proliferate.
"You haven’t seen anything yet," Blumberg said. "There is a flood of data coming."
Additional SXSW coverage:
SXSW: Instagram founders recap on ads, authenticity with Facebook integration
SXSW: Instagram founders critical of sweeping antitrust regulation in tech
SXSW: 'Enormous' opportunity for bankers willing to embrace fintech
SXSW: Media, entertainment, enterprise, health, policy, and AI to bind them all