For decades the VC world has been captured by an anthropomorphized notion of artificial intelligence. Much of this was aided by the ways in which AI innovators represented their technologies—for example, AI designs that intentionally looked like humans or animals when there was no technological reason for that design. By making AI seem akin to human general intelligence, innovators kindled the collective excitement of investors, journalists, and others keen to see humanity move further into a kind of “sci-fi” future. Rather than imagining AI improve street surveillance or security through a complex network of independent cameras, it was much more exciting to imagine a robot standing on a street corner.
Artificial intelligence is, however, very different from “general intelligence” that humans have. Rather than the cyborgs of science fiction, it deals with a series of “if, then” statements and pattern recognition technology to create an ‘impression’ of intelligence. By providing an AI with sufficient examples of “images of cats”, for example, an AI can distinguish those images from others simply through image and data matching.
An overly ambitious notion of what AI is capable of is partly why a 2017 study survey of CEOs found that AI ambition had far exceeded its practical usefulness. Still, VCs invested billions of dollars in AI machines that could mimic human motor skills (e.g. the failed robot Baxter) by flipping burgers, and most of these companies eventually failed. The almost cyclical nature of AI investment led to a series of “AI Winters”, where funding dried up and skepticism prevailed. The VC market seems to be reemerging from a recent “Winter”; AI startups took in $18.457 billion in VC funding in 2019. A key difference this time, however, is our idea of what role AI can play has transformed
Here are some of the major takeaways about what we can expect from this new wave of AI innovation:
More mundane, but also more helpful
- Previous innovations in AI seemed to bring us closer to the kind of general intelligence that could help automate away a large chunk of current jobs. The most recent wave of AI investment is focused on more practical applications and less ambitious applications of artificial intelligence.
- This includes analysis of massive data sets, supporting due diligence for law firms, photo recognition through data matching, and using image matching to help identify malignant tumors
- Examples: Viz.ai, Inc, Grammarly, Samsara
AI will change the nature of work, not end it
- AI is often conceptualized as a force replacing humans, but it has most productively manifest as a tool to be used by humans. As a result, AI innovations are changing the nature of work, but not reducing the need for it. A McKinsey report, which analyzed over 2,000 work activities across 800 occupations, found that about half of all the tasks people perform at work have the potential to be automated but only five percent of all occupations could be fully (100 percent) automated. These findings are also supported by similar studies conducted by the OECD and Brookings Institution among others.
- The most productive AI and the startups that are garnering the greatest investment primarily focus on aiding human labor rather than replacing it. This increasingly takes the form of automating tasks that require analysis of large datasets, rather than automating more complex tasks or work that requires motor skills (see also: Moravec’s Paradox)
The skills valued most
- The rise of AI innovation also has implications for both 1.) the skills founders of new technology companies value most and also 2.) for VCs looking to discern the industries that are most likely to experience significant innovation-growth now and in the coming years. Routine work in production, manufacturing, and food preparation, for example, are in the midst upheavals.
- More complicated jobs that require a broader array of tasks and require social skills—in essence, the skill of being a human—are unlikely to be automated. Workers who know how to use new AI technology effectively are also highly valued. Despite the fact that manufacturing has seen rising rates of automation, there is currently a shortage of workers who have the skills to manage the new machines. A challenge for new AI companies continues to be both consumer and worker adoption barriers as well as regulatory obstruction.