AiGg AMA salon recap pt. 3

The extended AiGg team - along with our wonderful advisors - held an Ask Me Anything (AMA) Salon to address questions from our participants, and it was a wide-ranging conversation.

We’ll cover the notion of data and AI readiness and environmental impacts in this third recap post.

Thank you to Brady Benware, GPT Now, and Leon Yeh, AiGg for steering the conversation! (And the rest of the team for jumping in!)

Q. John asked:

What's the bigger hurdle to AI adoption, fear of IP contamination, organizational fear relative to the unknown and the fact that things could get out of hand? Or is it the speed of AI? Or organizations not having their data ready for the effective and value of AI?

What’s the biggest hurdle to AI Adoption?

A. Brady Benware:

Yeah, I I think there's the challenge of data readiness, but there's also the relative inexperience of just how to work with AI. And I think that's something that everybody is facing. It's it's so brand new that there's just a level of inexperience - exactly how to use this new set of tools effectively?

And I think there is a a component of this that it's it's a reliability question. These things have to work at a small scale before they work at a large scale and and data readiness is definitely something that's going to be critical at a large scale. But you need the use cases to demonstrate the value to make it all worth it.

I think it’s going to take time for use cases and people’s comfort to evolve and deal with some of the things that we get hung up on in terms of IP contamination and whatnot. But I really do think it is a readiness in the ability to get value out of AI.

Exploring the misunderstandings around AI

A. Leon Yeh:  

I want to add, I think there's a lot of miscommunication about AI. I think a lot of thing people kind of take machine learning, neural networks and then they take chatbots like ChatGPT which is the latest AI trend, and people doesn't really understand what does AI even mean?

I'm a machine learning professional and a data scientist and I’ve been learning this as fast as I can… and progress on AI keeps moving so fast. Faster than my comprehension, even with my experience. So I can understand from people looking in into this, it could be just just a lot of misunderstanding about what AI does.

The way I look at AI is as a predictive machine. The thing that we should remember is AI is an indeterministic machine, meaning that it it's never answered the same question with the same answer, and that's very, very different than what we’re used to with computers. So, like a a dice that you keep on throwing… But it's so smart it's it's kind of answer almost 99.99% of the time exactly as you expected?

But over time you start seeing that it can start answering in totally unpredictable ways, and a lot of people hasn't really seen it do that yet. And scaling to the next level requires more understanding about the tools itself, yeah.

AI at what environmental cost?

Q. Julia asked:

I know ChatGPT has begun memory testing with some segments of its users. Gemini 1.5 now with its million tokens, driving increasing demands for computing power, and we're seeing exponential growth across AI driven applications. What are the panel members thoughts about the downstream environmental impacts free use while multibillion / trillion dollar companies controlling the market has broader implications, how do we avoid these costs being passed on?

A. Brady Benware:

I’ like to just offer an opinion here. I used to work in the semiconductor industry and and so very well aware of kind of the underlying technology behind all of the computation. And I think it's it's worth understanding what's going on in that underlying compute technology.

Does anybody remember the Cray supercomputer? It's a name that sticks in my mind… Distinctly. Today's laptops have about 100,000 times more compute per watt than the Cray supercomputer. (Editor’s note: See the Cray | iPhone comparison here.)

Notably it's probably taken about 40 years to get those gains, but we are continuing to improve computation per watt exponentially as we go. And one of the drivers for that computation per watt going forward, is that AI is helping us to design those next generation of compute chips.

And so you know, for the same one inch square of silicon that's doing the computation that's needed today and what looks like, you know, an exponential adoption of all of this, we're going to have much more efficient compute as we go forward. And largely it's going to be driven by the fact that we have this technology to to develop that next generation of compute.

So we're just at the very beginning of having the right computation to align to the computations that are even being done, and and so I'm very optimistic that we're going to be able to achieve this without really growing the footprint of that power consumption.

It's a little bit different than crypto because that was an unbounded. The more compute you had, the more computation you were going to do and and so that was a really bad situation. But in this case, yeah, I think we're going to be able to keep up with that demand and actually keep power consumption very normalized as we go.

Kathy Long-Holland:

Brady, I'm not as optimistic as you just given the overall demands… I hear you on the solution and there's some pretty innovative entrepreneurs working on that. But my worry is the demand is going to outstrip our ability. Now you think of Geoffrey Moore and the fact that it seemed to be ahead of us a bit. I'm not sure that's the case now, but it'll be interesting to see…. (Editor’s note: Geoffrey Moore - Author of Crossing the Chasm - about technology transitioning from early adopters to the mainstream market by jumping the innovation-adoption gap.)

Tara Smith:

I agree. I think it's going to put a strain on infrastructure in a way that we don't quite understand. I do think companies need to be mindful, but I do agree with the point that companies are also making advancements in cooling technologies. There's actually an interesting company in Seattle that can go out and find existing capacity on a grid to make it more efficient. It's called Smart Wires. So you know, technology is progressing on all fronts, but I think the downstream impact is severe and we do need to be paying attention.

Kathy Long-Holland:  

Yes, in a meaningful way, and I think we haven't done the investment in infrastructure all along. So we're starting from kind of behind the 8 ball. 

Tara Smith: 

With the hockey stick now. 

Kathy Long-Holland:

Yeah, exactly. Yeah. I think we need to be cognizant of that as we see brownouts potentially in our own, you know.

Susan Towers:

I think you see this with any new technology. If you look at the the impact of building electric cars right now, I mean, it's kind of incredible. Everybody thinks electric car is is an advantage over gas, but the cost to build the cars themselves, the weight, everything else has got a huge environmental impact. But you know, I think this is what innovation will take. There will be interim periods where there are challenges and hopefully we'll overcome them and get better.

Dru Martin:

Just observing how back to what Brady was talking about in terms of computing power, there's talk about starting to move AI computation to mobile. And if you think about the challenge there in terms of the computation that's needed and trying to place AI on something that has a relatively minor impact in terms of electric to use, that's going to be a huge driver in terms of the efficiency, because that's it seems like that's kind of the direction that Apple at least is going in terms of their technology with AI. (Editor’s note: relatively quiet Apple’s been in the news recently about the significant investments in AI. More soon.)

To connect with Brady at GPT Now, explore his website here. To talk with Leon, ping us here! More with Kathy, Dru, Tara and the rest of the AiGg gang in future recaps.

Resources from AIGG on your AI Journey

Our team comprises legal experts, anthropologists, and experienced C-Level business leaders dedicated to guiding you in your journey to leverage AI in your organization, safely and responsibly, protecting your employees, your IP, your brand and your organization itself through good governance, and proper risk management.

We invite you to connect with us and delve into the wealth of resources we have on offer. From a complimentary data privacy resource, AI Tools Adoption Checklist to comprehensive Legal and Operational Issues Lists, and from HR Handbook policies to interactive workshops—equip your organization with the knowledge to harness AI's potential responsibly.

Reach out for more information and to begin the journey towards making AI work safely and advantageously for your organization. We are your partners as this next transformation unfolds.