Donato Riccio

You have more different architectures, like a mixture of experts that can just turn on a portion of your model instead of paying for the whole model.

And we have smarter thinking models that can think adaptively based on the difficulty of the query.

The thing is, we don't even know how to use the intelligence we already have right now, and most companies are really still experimenting to figure out what AI can do and how to deploy it.

So I say that the bottleneck right now is not that AI is not smart enough, but really the cost is what can transform every industry.

Yeah, that's a great question.

I think right now we're in a phase where companies are just throwing their eye at everything to see what sticks.

And honestly, I think that's pretty healthy, because that's how you figure out what works right.

You experiment.

You don't have all the answers from the beginning.

You have to just take risks, see how your products evolve.

Some fail and some succeed.

But I think this phase won't last forever.

So eventually, the experimentation phase ends, and you need real results in the company.

When a company announces an AI initiative or a significant AI spending, I'd want to ask a few questions.

First, is this solving a real problem?

It sounds obvious, but you'd be surprised about how often the answer is no.

Is AI addressing an actual business problem?

I'll give you a simple test.

Can this problem be solved without AI?

Sometimes the answer is yes, so simpler is better.