A Business Perspective on AI
Today: It turns out harnessing the power of AI is expensive. Isn’t it easier and cheaper to merely deal with users?
The Agenda 👇
There’s more power on the outside—except if you rely on machines
Investing in AI spares the bargaining with demanding users
Is the AI route sustainable or will the multitude always win?
What does it mean antitrust in both scenarios?
You’ve read this before—the one idea you need to keep in mind if you want to understand the Entrepreneurial Age is the following:
Now there’s more power outside than inside organizations.
Again, more precisely this power on the outside is that which is vested in us individuals—a multitude of individuals, all equipped with powerful computing devices and connected with one another through vast networks, both physical (telecommunication networks) and virtual (your social graph on Facebook).
The fact that the power is on the outside doesn’t mean that corporations become powerless. What it means is that they can only secure power if they harness what’s on the outside.
The thing, obviously, is that you can’t simply approach this multitude of networked individuals and take what they have or demand that they create value for you.
Instead, you need to effectively make a deal with those individuals whose power you want to harness: I’ll provide you with high, affordable quality at scale; in exchange, you’ll let me use your power as part of the multitude to sustain my business and eventually make profits.
It’s all an alliance: corporations and individuals are aligning their interests so as to create value together and split the resulting economic surplus.
The problem with this alliance is that, well, it requires bargaining. And sometimes the weakest party (the corporation) has to make concessions to the strongest one (individuals), typically under the form of lower prices, increased quality, more customization, a broader range of products, shorter delivery times.
And this is why corporations are so interested in artificial intelligence these days. If you have the algorithm and you can train it with the right dataset, then at some point you don’t need individuals as much as you did before: all the data you need is already stored on your servers, in quantities that are quite sufficient to keep training those algorithms. As such you don’t have to bargain with the demanding multitude anymore, which comes with a key advantage: now that you’ve recovered effective control over that power that was once located on the outside, you can start raising prices again, lengthening delivery times, lowering quality, tightening the range of products on offer.
In short: thanks to AI, you can become more of a predator, because you don’t need the multitude as much as before.
Again, this is why artificial intelligence is so fascinating for the corporate world. Humans can’t be bought, they can only be rented. But machines can certainly be bought—and so, yes, ideally let’s get rid of those users who are doing part of the work and replace them with machines so that those on the outside can become mere customers again.
This leads to various questions (and I’ll leave it at that for today):
First, will many corporations be ready to make the huge initial investment that it takes to train these machines so that they won’t have to rely on the power of individual users anymore?
The sunk costs for choosing AI over the multitude is rather high if this article is to be believed: The Billion Dollar AI Problem That Just Keeps Scaling (via Matt Clifford’s recent Thoughts in Between). And the returns to scale are lower than with traditional ‘software’ companies, as explained by Andreessen Horowitz’s Martin Casado and Matt Bornstein in this two-part series: The New Business of AI (and How It's Different From Traditional Software); Taming the Tail: Adventures in Improving AI Economics.
Second, even if some corporations do make that huge investment so that it becomes easier for them to embrace a predatory approach with their customers (no need to seal an alliance anymore), won’t they be beaten by more entrepreneurial contenders who will renounce machines and bet on strengthening that alliance instead?
One last question: What becomes of antitrust in both cases? In the former case (AI-powered firms beating multitude-powered firms), it’s all about making sure that artificial intelligence doesn’t become a barrier to entry that protects predatory firms.
In the latter case, it’s all about shaping the conditions so that the multitude of networked individuals can negotiate with powerful corporations on an equal footing—making sure that the surplus resulting from this valuable alliance is split in an equitable way.
What do you think? Anything I should read on the economics of AI and the tradeoff between relying on machines and relying on users?
If you’ve been forwarded this paid edition of European Straits, you should subscribe so as not to miss the next ones.
From Munich, Germany 🇩🇪
Nicolas
