Generative AI

It's important for the way that it transforms the creator as they develop and explore that imagery.

And then it's important for the ways that resulting imagery can shape other people's experiences and thinking.

And One of my concerns with the degenerative AI tools is that they're kind of a black box.

You know, you can feed them a prompt and But that is this completely, you know, language is not, it can never replace the picture of, you know, the thousand words just don't do the thing, right? And what comes out of that black box isn't from your imagination, right? There's a fundamental divorce there.

That creative chain of custody has been broken.

Everybody was catching up in a way.

But the one thing we couldn't do, and this is where we always had to have guardrails, right? Because we don't, you can't, To leverage generative AI, you automatically violate your NDA, right? And so we could do generic things about the company, but we had a very thick firewall.

And there's only so much we could rely on the AI at the time.

And it wasn't until...

Lots of music books spend a lot of time writing about things that aren't music in order to try and explain the music, whereas you've taken on this really difficult task of trying to express the thing that music is through another thing, which is a real, real challenge, which is, again, you said earlier, like, oh, writing about music was easy when it was for ID, but I think that that's actually a completely opposite sort of process than what you engage in with this.

In any case, coming back to this thing about machines not distancing you from your emotions, I find that really interesting to read 30 years later, almost, when we have the rise of generative AI and we have this...

I mean, you used the word, I think, hellscape earlier.

Certainly it feels like a hellscape of generative AI, particularly when it's applied by corporations to extract value from things that they've stolen.

How do you feel now? Do you think that there's a...

I mean, in some senses, generative AI is a point along a process that you've started describing in this, but it also feels to me like something different.

How do you feel about it?

I think it is different.

So drawing on his background in aircraft engineering, Stevens says scrutiny should match the consequences of failure.

He lists the five questions testers should answer Before recommending a release, ending with this one, wouldn't more testing materially change our decision? For generative AI, he says complete coverage is close to impossible, so teams need representative data sets, repeated evaluations, and some math around statistical uncertainty.

So if you want to learn more, definitely check it out using that link down below.

So another thing on LinkedIn that caught my attention is about this new open-source solution.

And I think...

If I go all the way back to the first time I saw anything, and by the way, we're going to talk a lot about different types of, quote, AI, generative AI for content creation, and then AI as in LLMs and using it for work and things like that.

But if I go back to when I first started being aware of it, I think it was probably the Will Smith eating spaghetti.

And I remember looking at that being like, oh, this is fascinating.

And then sure enough, everything just started getting better.

Everything started getting faster.

And I kind of saw it as like I experimented with generative AI like early, early on, like maybe I don't even remember.

It was like probably two years ago at this point.

And so for our Spring audience, Spring AI is really good at that.

It's a framework for building or integrating Gen AI capabilities into your applications.

And it can use the frontier models.

It can also use open source models.

Just give the audience an idea of some of the things that can be done in this new age of AI inside of an application.

You've opened up a whole bunch of new topics.

We've been talking about Gen AI and the Gen AI use case, but you really opened up, at least for me, all of these other ways that I can take the data that we all have, we're all surrounded by it, and we've been talking about this for a long time, and we kind of stopped talking about it as we rolled out of this rag phase.

But what you have done for me, at least, is open up the idea that I have all of this data, if I'm in an enterprise, How can I be using that to help my customers? How can I be using that to help the business? Prateek, this has been an absolute wonderful, the gem of a conversation.

And I registered on in MIT.

So I said, OK, I'm going to learn all about Gen AI.

I'm going to learn all about a Gentics AI, because if I want the data source, I still need to know what to do with it to provide personalized health care.

So it was two folders.

So I wanted to recreate Moab, but go even bigger.

And have those kinds of connections in your day-to-day work.

I'm curious what would change, what would be different if you could get everyone on your team or everyone that you work with, when you think about even cross-functionally, not just lockstep with you, but what if they were to come home and say, I need to do an agentic and gen AI deep dive with MIT? and had that kind of motivation and clarity that you had, what would change? What would be different if everyone around you had the same kind of energy, confidence, clarity that you gained from, from the experience?

I think it's giving ourself, I was just talking about it this morning, because one of my colleagues asked the same thing.

Yeah.

Generative AI is less problematic than like pictures, but it's still stealing information.

Sadly.

Exactly.