Frank Chen — In a few years, no investors are going to be looking for AI startups
Soon, AI will be taken more and more for granted as a foundational building block of technology. For future startups, that means that AI will almost certainly be essential to their solutions, and questions will be raised if it’s not.
But I don’t share the view that VCs should assume every startup will use the best available AI. We shouldn’t even be suggesting that there will be a “best available” at all. Actually, I think it’ll be even more important in the coming years for VCs to understand what it means to have AI when using it will be nearly ubiquitous among startups.
AI isn’t just-another-technology. It’s true that we should expect that AI among startups will become commonplace and integrated everywhere (hence we are investing in many “platform technology” companies and core components such as next-generation cameras). The foundational impact that the AI age will have on the way business is done cannot be understated. AI done well means that:
- Work that used to require expensive and non-standardized education can gradually become cheap and standardized — meaning the quality of work done can be consistently high and easily accessible. In healthcare, for example, AI will bring efficiency to things like drug discovery, patient monitoring and diagnostics, paving the way for precision medicine and freeing up doctors’ time time for less automatable tasks like attending to the personal and emotional needs of their patients.
- Tasks that used to be economically inefficient to do, will soon become commonplace and profitable — meaning significant value can be realized from what used to be “unsexy” or “uninteresting” businesses. On the one hand, large-scale robots will automate necessary tasks like sorting recycling (AMP Robotics) or harvesting produce (Abundant Robotics) that are expensive with human labor. On the other, AI will allow extreme customization of services to the individual level, which we’re already seeing with automated trading platforms like Wealthfront. This isn’t possible without inexpensive intelligence.
- The ability for us to learn, process, and understand the world will drastically improve through AI. Concepts & processes that were too complex to understand or remember will become solvable, everything from increasing traffic efficiency to understanding disease and developing treatments, will become drastically more achievable.
- Our ability to collect data, rapidly evaluate, iterate on our decision making, and see the effects of our new decisions will allow the learning of humanity to make leaps and bounds that were previously unimaginable.
But that presents the issue: we can’t assume that everyone will be using the best AI. That will be precisely the challenge for VCs in the near future.
It’s true that startups of the future won’t simply be able to say that they use AI. Statements like that just add to the hype. (I for one am very glad so many people in the community have already pointed out that problem.) Instead, VCs should expect that founders are able to explain why they’re using AI and why their approach is the best and most relevant for their use-case. That’s because AI is not one piece of technology, but rather a concept that describes many different methods for reaching new answers to complex problems with machines. VCs should also appreciate that not every problem can or should be answered with AI, and keep that in mind when assessing future startups.
Unlike developments in cloud storage and mobile, the way that AI impacts us will continue to grow and evolve in ways that we can only imagine right now. There isn’t a clear line that we can reach where we’ll be able to suddenly say, “We have AI now!”. Even the mere definition of intelligence is changing from both technological and biological perspectives.
That doesn’t mean having AI is a dead end. VCs will instead need to dig deeper into the tech. Change the questions to: does the specific AI model make sense for the problem? And: Is the training data reliable? VCs that don’t spend enough time examining the core technology because they assume it’s there will put themselves at risk. There won’t be one single “best available” AI; there will be many, with some being better suited than others depending on the problem. We all, both VCs and founders, also need to understand which problemswill or will not be solved by AI, and how to build businesses that can best make use of these new technology breakthroughs.
AI as a foundational technology will certainly become standard, but the execution, expertise, and techniques are a whole other story that can’t be taken lightly.
Let’s block ads! (Why?)