Are there other limitations in the use of AI in this context, in addition to a lack of available and appropriate data?
“Because these technologies require training on existing data, they are more likely to predict a start-up’s success accurately if historical data contains lots of comparables. The trade-off is that VC firms that adopt data technologies consequently tend to tilt their investments towards start-ups developing businesses similar to those
already explored. While they become better at avoiding failures within this pool, they are also less likely to pick
innovative start-ups achieving rare major successes.”
“The widespread adoption of AI in VC could raise concerns about innovation on a larger scale. It could lead to a skew in investments towards more established ideas, potentially side-lining ground-breaking innovations that don’t fit established patterns.”
Maxime Bonelli
“VC is disruptive by nature and using historical patterns to identify something truly innovative doesn’t make sense. What could make sense, however, is identifying what has gone wrong in the past and using that to raise red flags on future investments.”
Merrick McKay
“Current machine learning algorithms search for patterns, but by definition, breakthroughs do not fit previously established patterns, and this might be a concern for VC firms interested in using AI. However, I am optimistic that AI can help in the VC world by broadening deal sourcing, rather than relying on existing networks. Companies may be being passed over because their founders don’t fit the standard casting of VC-backed founders. They may not have attended the most prestigious schools, for example. That is where I see potential to shake up the deal-sourcing process.”
Léa Stern
“History may not repeat itself, but it often rhymes. I am doubtful AI will identify the next star, but it can help us look at probabilities and flag, say, when valuations are higher than historical trends.”
Joshua Lowe