Nova: Imagine it is the late 1800s and you are looking at a lightbulb for the first time. Most people saw a glowing glass orb. But the entrepreneurs who changed the world saw something else: they saw the cost of light dropping to near zero. When light becomes cheap, you do not just light your desk; you change how cities are built and how factories run. Joshua Gans argues that we are at that exact moment right now, but instead of light, the thing getting cheap is prediction.

Nova: Not quite a crystal ball, but definitely a math-heavy one. Joshua Gans, who is a heavy hitter in the economics of innovation at the University of Toronto, wrote The AI Entrepreneur to strip away the sci-fi hype. He says if you want to build a business in this era, you have to stop looking at AI as magic and start looking at it as a drop in the cost of information. Specifically, the cost of knowing what is likely to happen next.

Nova: That is exactly the puzzle Gans solves. He argues that when prediction becomes cheap, something else becomes incredibly expensive and valuable: human judgment. Today, we are going to break down his framework for how to actually build a company when the math is handled by machines, but the strategy is still very much up to us.

Nova: To understand Gans's perspective, we have to start with his core thesis: AI is a prediction machine. In economic terms, when the price of a foundational input like prediction drops, two things happen. First, we use more of it. Second, we start using it for things we never thought of as prediction problems before.

Nova: Think about a self-driving car. We used to think of driving as a series of complex physical maneuvers and human reflexes. Gans points out that a self-driving car is actually just a massive prediction exercise. The AI is constantly predicting: If a human were in this seat, would they turn the wheel five degrees to the left right now? It is predicting the human response to environmental data.

Nova: It means your business model should probably center on a prediction that used to be too expensive to make. Take Amazon. Right now, they use a 'shopping-then-shipping' model. You buy it, they ship it. But Gans suggests that as their AI gets better at predicting what you want, they could move to 'shipping-then-shopping.' They ship it to you before you even buy it, because their prediction is so cheap and accurate that the cost of you returning it is lower than the profit from the instant sale.

Nova: Exactly. He emphasizes that for an entrepreneur, the goal is not to have the best AI. The goal is to find the most valuable prediction that no one else is making. He talks about the 'Minimum Viable Prediction.' Instead of trying to build a general AI that knows everything, what is the one specific thing your customer needs to know to make a decision?

Nova: This brings us to the most important part of Gans's framework: the relationship between prediction and judgment. If AI handles the prediction, humans have to handle the judgment. Gans defines judgment as the process of determining the reward for a particular action in a particular environment.

Nova: The judgment is deciding whether you care about getting wet. The AI tells you the probability, but it cannot tell you the 'utility' or the value of the outcome. In a business context, an AI might predict that a certain customer is 90 percent likely to churn. The judgment is deciding: Is this customer worth saving? Do we offer them a discount? Or do we let them go because they are actually a low-value, high-maintenance client?

Nova: Precisely. And Gans argues that as AI gets better, the demand for great judgment actually goes up. This is a huge opportunity for entrepreneurs. You do not necessarily need to be a data scientist; you need to be a 'judgment specialist.' You need to understand your industry so well that you know exactly which predictions are worth acting on.

Nova: That is the scaling challenge of the AI era. Gans suggests that successful AI entrepreneurs are those who can codify their judgment. They create systems where the AI's predictions are automatically funneled into specific actions based on pre-set human values. You are essentially 'programming' your judgment into the business's DNA.

Nova: That is the risk. Gans calls this 'automated stupidity.' If your judgment framework is flawed, a high-speed prediction machine will just scale your mistakes. This is why he advocates for a tight feedback loop. You need to constantly check if the predictions the AI is making are actually leading to the outcomes you valued in your judgment phase.

Nova: Now, let us talk strategy. Gans introduces something called the 'Entrepreneurial Strategy Compass,' which he has adapted specifically for the AI world. He says every AI startup faces a fork in the road: Do you compete with the big guys, or do you cooperate with them?

Nova: True, but Gans points out that competing is incredibly hard because of the 'data moat.' Big tech companies have more data, which means better predictions. So, if you are a startup, you have to choose one of four quadrants on his compass: Intellectual Property, Architectural, Value Chain, or Disruptive.

Nova: If you go the Intellectual Property route, you focus on the algorithm. You patent your specific way of analyzing images and maybe you license it to hospitals or big medical tech firms. You are cooperating with the existing players by giving them a better 'prediction component.'

Nova: Then you look at the Disruptive quadrant. This is where you use AI to serve a customer that the big guys are ignoring. Maybe you create a cheap, AI-powered skin check app for people in developing countries who do not have access to dermatologists. The big medical companies do not care about that market yet, so you can build your data moat there without them noticing.

Nova: That is the boldest one. That is when you use AI to completely change how a value is delivered. Instead of just helping doctors, you create a whole new system where the AI is the primary point of contact, and the doctor only gets involved in the 1 percent of cases the AI cannot handle. You are redesigning the entire architecture of the industry.

Nova: He does. He notes that architectural changes require 'AI-First' thinking. Most incumbents—the big, established companies—cannot do this because they are too tied to their old ways of working. They try to 'bolt on' AI to their existing processes. A startup's biggest advantage is that they can build the process around the AI from day one.

Nova: This leads to a fascinating point Gans makes about why big companies often fail at AI despite having all the money and data. He calls it the 'Incumbent's Curse.' It is not that they are lazy; it is that their existing systems are too successful to change.

Nova: Exactly. Gans points out that for an incumbent, switching to an AI-first model often means destroying their current value. If an AI can predict creditworthiness perfectly, the bank might not need the complex branch network they spent billions building. So, they resist. They use AI to make the loan officer 5 percent faster, while a startup uses AI to make the loan officer obsolete.

Nova: Gans has a great answer for this: 'Strategic Experimentation.' He suggests that AI entrepreneurs should not wait for a perfect dataset. Instead, they should design their initial product specifically to generate the data they need. It is a 'data-trap' strategy.

Nova: Precisely. You offer a service that might even be manual or 'human-in-the-loop' at first, just to collect the specific data points that will eventually train your AI. Gans emphasizes that the 'entrepreneurial' part of AI entrepreneurship is figuring out how to get that first 1,000 data points when no one knows who you are.

Nova: That is a huge technical and strategic risk. He advises entrepreneurs to always keep a 'hold-out' set of real-world data or to constantly introduce 'noise' and new experiments to keep the AI from getting stuck in a rut. It is about being a scientist as much as a CEO.

Nova: We have covered a lot of ground today. From Joshua Gans's view of AI as 'cheap prediction' to the vital role of human judgment, and the strategic compass that helps startups navigate the land of giants. The big takeaway from The AI Entrepreneur is that the technology is the easy part—or at least, the part that is becoming a commodity. The hard part is the strategy.

Nova: That is exactly what Gans wants us to realize. AI does not change the laws of economics; it just changes the variables. If you can identify a high-value judgment and pair it with a low-cost prediction, you have the foundation of a powerhouse company. But you have to be willing to experiment, to fail, and to look at data not just as numbers, but as the fuel for your strategic vision.

Nova: That is the spirit. As Gans says, the future belongs to those who can see the predictions coming and have the judgment to know what to do with them. Thank you for diving into this with me today.

Nova: This is Aibrary. Congratulations on your growth!