Freeing up Human Brainpower
I am personally very excited about AI being an enabler in innovation. This is because it is going to replace so much of the repetitive tasks in our work. Therefore it will free up a lot of human brainpower and mindshare to do more innovative things. I think it is going to make certain products that are already commercialized today even cheaper. Because AI will optimize the supply chain, production, distribution of products and services. But at the same time, the most authentic human experiences are actually going to command a much greater premium. What today costs a thousand dollars in terms of authentic experiences of artists, designers, or creative solutions will cost ten times more in the future.
AI Management
Ultimately this is going to change the skill set for tomorrow. When I look at data scientists on my team. They are much better managers than myself when I was there age. Because working with AI fundamentally is amount management. It is what task you choose to do yourself and what task you give machines to do. It is less about doing one task and it is about thinking about how you scale yourself. How to delegate and how to design a system that has many different algorithms working for you. So I think the next generation will breathe and live the AI. Therefore it will be very intuitive in terms of management and they probably don’t need MBAs anymore.
Startups don’t need AI right away
There is a big temptation to use AI everywhere in our products and services. There are certain situations when you are a new startup, you actually don’t need a neural network to solve. If anything it will be way too costly. In the beginning, you will find a simple algorithm even like a linear query can be quite effective. Not to say that AI wouldn’t be as effective but as a first step, there are lower cost ways to highlight an AI-based solution.
AI Challenges
I think number one, the biggest challenge is finding the right problem to solve with AI. Then finding the right stage for your growth to leverage AI.
User Experience
The second one is about the experience. Today we personalize so many different experiences when you open your Netflix when you open your Facebook and all these algorithms have AI behind them. But there needs to be the right balance of giving you just the right amount of experience you want without being creepy. It needs to be intuitive and fluid and comfortable. That is something we need to think about a lot when it comes to building AI products. How do we make the this comfortable for the consumer in their lives vs making it disruptive?
Humans working with machines
The third thing is how humans will work with machines. For humans to work with humans is hard enough. For humans to work with machines is going to be harder because the machine does not speak the language, you can’t go get a beer after work with your machine learning algorithm. So you have to think of everything on a mathematical level, setting up the objective function, designing what is a good outcome vs a bad outcome, and these are things that will be very important going forward. AI works in mathematical terms, so you have to express every single decision you make or you are going to make in math. It actually forces us as humans to think about these problems that were not real before but now we design these intelligent assets or intelligent vehicles that are moving around, you have to express some kind of trade-off mathematically and that is something the industry is still figuring out.
