Part 2: AI and the Economy: Transformation and Disruption

Where we are in the arc

The moderator, Richard Lui, introduced Jason Alan Snyder as Chief AI Officer of Momentum Worldwide and asked what brought him to AI and where we are in the arc. Jason said:

Sure. I'm a technologist, a futurist and an inventor. I've spent the majority of my career at the intersection of culture, technology and business. My background is in systems, deep systems engineering and embedded computer systems. So I've been inside of machine learning and AI for a very long time.

To answer the question about where we are in the curve, I tell the story this way. I think it's apt that we're having the conversation here in Los Angeles, so close to Hollywood. Because we have people that aren't lawyers and they're not doctors and they're not software engineers, but they play them on TV and in the movies. And I think that's where we are with AI today. Generative AI doesn't reason; it performs reasoning, in the same way an actor would perform a part.

I think that's very important to understand, because I don't really believe we have artificial intelligence today. I think we have artificial inference. We have machines that are very good at predicting language. They have good manners and they are very, very good at rhetoric. And where that comes into play with business, obviously, is that whoever is closest to the consumer controls the conversation.

The size of the prize

The moderator cited an estimate that AI will contribute 20 trillion dollars to the economy, another panelist put the figure closer to 4.4 trillion, and the moderator asked about the cost of implementing it. Jason said:

Yeah. There are many domains inside of it. So automation is one part of it. But AI generally is a combinatorial innovation. You have massive parts of this which are data, and managing data decay, and the information that's feeding the models. There's the model creation. And then there are all of the new and emerging technologies which are going to contribute to advances across AI beyond the chipsets, things like quantum and biological computing, that are going to change the dynamics. So I think it's a very complex environment with a lot of deltas.

Asked to pick five trillion or twenty. Jason said:

I'd probably say it's closer to 20.

Fake truth

After a discussion of the supply shock AI brings to jobs and to capitalism, the moderator asked the panel for quick reactions. Jason said:

I think the only other point I would add is that we're at a place where our technology has surpassed our legislation and our morality. And we are entering new territory where generative AI is actually creating a new understanding of our reality, some of which is based on fabricated information. So it is no longer fake news. It is kind of fake truth. And we have to deal with this complexity in ways that we haven't before, and it's going to take us a minute to unwind and understand what this means.

Ahead or behind China, and the data problem

The moderator asked each panelist whether the United States is ahead of or behind China. Jason said:

Ahead today, for sure. But I think the data point is so important, because without the data, these systems don't work. They're meaningless. And we have tremendous data issues. Whether it's about consent and how that data is collected. Whether it's about how that data is decaying and then being hallucinated and amplified in negative ways inside of these models. And we really need to start to codify practices, really from a business perspective. Because we are focused on innovation, not regulation. And so it's going to be the responsibility of everyone in this room and everyone endeavoring inside of this work to make sure that we are maniacally focused on making sure we have the best possible data that's based on consent.

The moderator asked: if we need 100 in terms of data, do we have 50, 40, 75? Jason said:

It's difficult to answer that question, because the veracity of the data that we have is unclear.

Pressed for a general sense of how empty the glass is. Jason said:

It's pretty empty, in my opinion. And the real risk that we have is the synthetic data that's being created by these models. When we're looking at the big tech companies, it's the research that they're doing which is being largely automated and largely using synthetic data based on training data which is bad.

Chips, and the bottleneck

Asked to react to the hardware stack, the chip, and the data centers. Jason said:

Sure. I think with chips, chips are just one way of solving the problem. Quantum computing is a thing, biological computing, protein based computing. When we have good brain computer interface, there'll be shared compute inside of that. So I think that chips for right now are just a moment in time, and I think we need to remember that with the exponential growth of this technology, we're going to see new ways of solving this problem computationally.

Asked for the bottleneck today and in five years, in one word. Jason said:

Ethics.

Closing

Asked for a closing statement. Jason said:

If the service is free, you're the product. Be careful what you consent to.