How do Venture Capitalists Evaluate Artificial Intelligence Companies?

Advanced Complex Systems, one of the most unique courses in the graduate curriculum, introduces and explores a cluster of related concepts that pertain to systems in which large numbers of “agents” — people, companies, or any other kind of autonomous entity — interact. Included in these concepts is the dynamical systems theory, which introduces the notion of “attractors”: states or types of behavior toward which a system is naturally drawn.

#systemdynamics: Recognize the role of attractors and sensitivity to varying conditions in the behavior of complex systems

Through his research on the past 50 years of technology investing, Veiga da Silva documented a shift away from funding models driven by banks, individuals, and private equity firms, and toward a model driven by venture capital firms. Based on his analysis of trends in AI, he hypothesized that the rise in enabling infrastructure — especially computing power and large datasets, in short, big data — might have something to do with this shift. If true, this infrastructure hypothesis may have interesting implications for other industries. Veiga da Silva wondered: Are there other industries for which the enabling infrastructure is presently being developed? What are they? And who will benefit?

It can be tempting to draw grand conclusions from a few salient observations, rather than collect data from a subset of individuals — a sample — to base your inferences and observations. Veiga da Silva was sensitive to this temptation for two reasons. First, the modern AI industry is dynamic, turbulent, and colored by an enthusiasm that may draw people’s attention to good stories rather than durable truths. Second, venture capitalists have reasons to be secretive: they are in competition with each other. What they reveal to the public about their methods, they also reveal to their competitors.

#sampling: Design effective sampling methods and evaluate the interpretation of results accordingly

Moreover, he provides the caveat that the published studies of venture capital decision-making included interviews with only 200 or so subjects over the past 40 years. Since the industry has surely employed thousands of people at that time, Veiga da Silva was cautious in his estimation that he understood how it actually works.

As he continued his research, Veiga da Silva noticed other reasons for careful analysis. AI is so scientifically sophisticated that many venture capitalists are not well-equipped to understand the companies they review. Those companies are aware of this gap in expertise and, in some cases, attempt to present themselves as doing more than they actually are, in order to attract investors’ interest in an emerging high-tech space. Ultimately, this not only hurts the investor but the market overall.

#gapanalysis: Identify gaps (in knowledge, in market offerings, in a range of ideas, etc.) that reveal where a creative solution is required

In an effort to identify and call attention to this gap, Veiga da Silva wrote that what the market needs is a set of clear criteria that both entrepreneurs and investors can use to maximize robust and reliable decisions, pushing the frontier of successful AI solutions, while minimizing wasted money and time. One of the venture capital firms profiled by Veiga da Silva publishes its criteria. Among the 17 criteria, there are factors related to the uniqueness of the product or service, its projected reception in the market, and its defensibility from the competition. Being transparent about their methods yields many benefits to these investors and the industry overall, including the ability to see what is missing.

#fairness: Recognize and work to mitigate unfair outcomes and practices

In the case he profiles, Veiga da Silva observes a poignant criterion that was not included: an assessment of ethical principles of investing, including assurances that companies run by women and minorities will not suffer competitive disadvantages. We know, he writes, that such discrimination is indeed a problem in the high-tech industry (see, for example, Liza Mundy's April 2017 piece in The Atlantic), and so we should be taking steps to avoid it.

Delving deeply into an industry that he knew little about required Veiga da Silva to think both creatively and analytically. Documenting the interplay between different types of organizations and processes, such as governments, corporations, and entrepreneurs, required the application of concepts from complexity analysis and system dynamics. Knowing how much to generalize from this industry to others, or from one venture capital firm to another, required knowledge of statistics, and identifying fruitful directions required problem-solving skills and ethical reasoning. As a corporate attorney, Veiga da Silva had a solid foundation to build upon; the skills he learned at Minerva, and the professional contacts he made in the course of his thesis research, are helping him understand and move into a new field.

Apply to join the Master of Science in Applied Analyses and Decision Making!