Mario Harik

And I think currently where AI is and where robots are, they're still not there in terms of being able to perceive what human consciousness is to a certain extent.

So going back to your question, it's tough for me to tell, is it gonna be 10 years before we have effective humanoid robots?

Is it gonna be two or is it gonna be 20?

Because the rate of change is very fast, but at the same time, the problem to solve is very, very hard as well.

Now, more broadly though, if you look at AI, it's truly remarkable what has happened over the last two, three decades.

I remember I built my first AI neural network back in the late 90s.

At the time, it was running on a Pentium CPU that was, from a capacity perspective, was capable of at the peak of maybe one gigaflop in terms of compute power.

You look at today where the latest Google TPUs and where the NVIDIA chipsets are at, we're talking five beta flops.

That effectively is five million times more powerful from a compute perspective what existed two to three decades ago.

So when you think about the

how large the AI models were at the time, if you have something that has 10 billion parameters, you're pushing the envelope.

Today, you're dealing with two, three, four, five trillion parameters in terms of how these large language models operate.

So I think if I roll that forward, the trajectory has been effectively very, very fast or very steep in

And if that continues, we will see how that materializes into robotics.

But I think in the near term, I think it's more purpose-built robots that are using AI to try to automate a certain task and make effectively the outcomes more efficient accordingly.

I think it started with my parents.

My dad was an entrepreneur and he's also an engineer and was also an amazing business person as well.

Very wise person.

If you meet him, you just spend 15 minutes with him, you say, okay, that person is very, very, very wise.

And I saw since we were kids, the work ethic that he had.