FIR 92: AI-AI-O...oh oh !!


Manage episode 270039305 series 1410522
著作 Grant Larsen の情報はPlayer FM及びコミュニティによって発見されました。著作権は出版社によって所持されます。そして、番組のオーディオは、その出版社のサーバから直接にストリーミングされます。Player FMで購読ボタンをタップし、更新できて、または他のポッドキャストアプリにフィードのURLを貼り付けます。

AI-AI-O...oh oh!! What are the real AI problems and the threats to your business, or even to YOU!?? Should I not use AI in my business?

What does that an AI problem or threat look like? So, to get started with this, I looked at what several organizations were saying on this. Here's one this came from an article out of tutorials point they were claiming that AI is growing at such a rapid pace. It seems magical magical is the term they indicated the researchers note that AI could take over human jobs. Alright, there's certainly some truth to that. I don't know about the magical part. Probably because while I'm in AI, and I build AI stuff and doesn't feel so magical, but Okay, here's some threats. All right. So there's one threat to privacy mail to privacy, right privacy of humans, right. So an AI program that recognizes your speech and understands your language is theoretically capable of understanding your conversations on emails and telephones, right. Okay. So that's it.

Here's another AI Oh Oh, it has to do with the threat to human dignity. What that means is some feel that AI systems will start replacing human beings in certain industries. All right, so that's a that's a threat or problem caused by AI. Here's another one a threat to safety. So that they feel that when an AI system is self improving, it can become so mighty that humans will have a hard time stopping it. Okay, so that's what one group claim that came out of it tutorials point article. Here's what some folks on Wikipedia, were articulating. They were talking about the AI control problem. So the concern here basically is, sometime in the future, the term that's used is there will be a super intelligence and the super intelligence will overcome human intelligence. Now, while this is a potential with the AI that I've been working with and seeing, I believe

This is a long ways off from the moment of recording this now brilliant minds like Stephen Hawking have suggested that, you know, we need to start figuring out now how will contain that super intelligence AI system in the future. So that's fair. But for right now keeping your AI under control that's in the minds of the great thinkers. But let's talk about the AI problems that are actually a little more realistic in terms of here. And now because, well, you're solving business problems today. And what do we know? What does that look like? So one of those is the problem of perverse instantiation lines and tigers and bears, oh, my perverse instantiation who comes up with these names? All right. So that basically says, hey, we've got a problem where the AI perhaps gets assigned the wrong goals by accident, right. And what it means is that the smarter and more capable an AI system is, the more likely it could find an unintended shortcut that satisfies its goals.

So to say it another way, the end does not always justify the means, again, today, we're not quite there, right? It's not quite that far along. Now a human could take the results of some of the AI stuff and use it in matters that might be unethical. But is that really the AI system doing that? Or is it the human? So I'll give it I'll give an example of this going to the extreme though. It's kind of like that will smith movie called iRobot? I don't know that's made eight or 10 years ago, I don't remember but, but it's where this AI system determines that running an efficient society is the most important thing and given given that the humans are inefficient, well, then the humans need to be constrained. Right? And so you know, watch out AI alert. So that's an AI AI Oh Oh, in any event, today, I don't see that yet as an issue.

Let me turn to an article this came from McKinsey, they're a great research and consulting organization. They gave this article on what are some of the problems and challenges around AI? And one of them pointed out that, and I think this is very relevant today, it's around the explain ability of AI, right? Sometimes it's like, Oh, it's such a mystery that you just can't understand what the heck this is. I'll give an example of it, right? So when you have a collection of AI algorithms, and at the end of the day, it's software running math, it's looking at probabilities. Deep breath. All right, that's what it is. And when a bunch of them, you know, come together and say, hey, there's high probability that this is the direction to go. It's not always black and white. And sometimes as a business owner, he'd kind of like to know is it a yes or is it a no, right?

Should I do this or should I do it not and so sometimes the explainability as a little tough. It's like, well, you should do it only if it's a Tuesday in Belgium and it's raining in Pennsylvania, right? In fact, not long ago, I was doing some AI for an organization and had an answer kind of like that. It's like, hey, when it's raining, wherever your customer is sitting, they tend to be indoors more, and they're listening to what you're talking about in selling, therefore, your sales go up. And so that kind of thing actually happens. So this is actually where I stand. I stand with McKenzie, because they go on to start describing what the business benefits are of using AI. So I believe that we can make the predictions from AI understandable for small to medium businesses, even though AI can have an explainability problem. Lots of times that is incumbent on us as the as the humans to do that.

So I intend to democratize AI for the small to medium business owner. Now what does that mean? It means that we can apply AI today to small to medium businesses.

It can provide value to grow your business. So let's look at some practical AI use cases. This, this next piece also comes from, from a McKinsey article, they're talking about how there's a set of practical AI use cases that are found across in their words, all sectors of the economy and multiple business functions, from marketing to supply chain to operations. Deep breath, Okay, that makes sense. All right. So you can use it in sales, marketing, supply chain, improvement of your operation efficiencies, and in many cases, the deep learning techniques are used. So what's deep learning deep learning is kind of I'll give an example of deep learning. Let's say you're standing in a room and you need to leave the room. Okay, as a human, we will move around and look for the door. So that's the first problem find the door and then we will engage our legs if you will, to move towards the door.

Then we'll grab the handle, then we'll walk out the door and then we'll turn and decide which way to go once we leave it. That's a combination of problems stacked one after another. And you could have a neural network for each of those steps, right? Find the door neural network. Whoa, what does the door look like? Okay. All right, move towards it. Okay. Have I arrived at the door? Yeah. All right. Okay, now that I've, you know, exited the door, for example, which way do I go? so deep learning is nothing more than taking a collection of those problems, stacking them one next to another, and having all of the math and probabilities that are needed to help you answer those questions. So deep learning, okay, make sense? You can actually use that stuff today for small to medium businesses. But as a business owner, you don't have to worry about that part. Well, you want to think about is, what are my use cases? Am I trying to improve my sales or reduce my refunds right or improve my my supply chain?

So here's from the McKinsey work, they said look and their analysis of for over 400 use cases across 19 Industries, they found that AI improved on traditional analytics techniques in 70% of those cases. All right. So that's critical to know that, all right, the research also showed that using these deep learning techniques, they can generate as much as 40% of the total value of all analytics techniques by 2030. So in other words, the more that we can stack deep learning techniques, or I'll just call it artificial intelligence, for now, at least for this conversation, it's actually going to have a dramatic improvement over just core analytics techniques. The other thing that they estimated is that there will be a value to these techniques of up to $6 trillion in the GDP. All right, that's that's a big number.

But you know, what's interesting is even with those big numbers. Still they go on to say that only 21% of organizations have actually embedded AI into their business at some level, only 21%. And so the key here is, if you feel like you're behind the times when you're not, right, lots of organizations still have a ways to go with that. And in fact, most of these have been larger organizations. So for the small to medium business, the opportunity is ripe for you to actually leverage and take advantage of AI and use it to leapfrog your complex, you know, your competition. All right. So I wanted to point out something on the on that research that they stated, this is kind of interesting. They said, there's their surveys, surveys show that early AI adopters tend to think about technologies expansively. In other words, they're focusing more on how do they grow their business and increase their market share, rather than their rather than companies.

Focusing on reducing costs. Now, that's not saying don't reduce costs. But if we're fixated just on that part of it, then that actually tends to reduce the opportunities for us to leverage AI to actually grow the business. And they're finding that the early adopters tend to be more of the mindset, hey, I want to grow and expand and therefore AI becomes a mechanism that helps them to move towards that. The other key thing that that sort of jumped out in this was, they stated that highly digitized companies tend to invest more in AI and drive greater value. So another way of saying that is, hey, if you got an online business, then you're a perfect candidate for this. That's not to say non online businesses can benefit. Certainly they can. But they just saw they said higher value for the online businesses, at least for today, right with the current state of where AI is. But wait, are we talking about AI problems here? I got a little excited on some of the benefits, I can go back to pummeling AI, let me pummel AI.

So this comes from another research article out of Oxford. It was a study where they said that more than 47% of American jobs will be under threat due to automation By the mid 2030s. Now, I need to say here, okay, that's automation. Not all automation means AI. I mean, you know, a car is some form of automation, right? So, what I will point out, though, is that they do go on to state that AI though they so they narrow it down. And this is per the World Economic Forum, a research they did said that 75 million jobs will be replaced by 2022. Whoo. That's, that's a lot. I think. I think that's a little high, man. That's like in two years, we'll see. Now, the thing that's kind of interesting about this is that most of them come back and say this

The kinds of jobs that will be replaced will be the low skill, low educated jobs. And so what it does is it says over the next 10 to 15 years 40% of the world jobs will be replaced by AI based bots in the next 10 to 15 years. 40%. Right? That's sort of venture capitalist, his name is Kai Fuli says that like, Okay, if it's 40%, and they go on to again, they go on to declare, hey, this is for organizations or this is for skill sets that are, you know, that are the lower skill sets, then what it does is it gives us time to address the education gap we have about a decade right, some jobs will already start changing. But I also want to baseline on something that even today, there are not many horse carriage drivers and maintenance teams, so the workforce can and should evolve. So it's I don't I don't want this to be Oh, all gloom and doom because in reality, we have been retraining the workforce for generations as new rounds of technology and capabilities come, so this was really a call to while some view it as a threat.

I think as a business owner, we should really look at what's the impact to our workforce, if I could leverage AI to take care of these sort of activities, most likely it will bring productivity gains to my business, and quite frankly, today at the current AI maturity levels, that's actually how AI helps the businesses today, it's not some super intelligence that's, you know, running robots around and saying, you know, humans go away. It's not that at all, it's much more, oh, I'm gaining efficiency improvements. I'm getting insights into where I should be spending my resources and my time, and therefore I get these productivity gains to the business and that's really the benefit and the value of AI today. Okay, so there's tons of myths out there. There's tons of things that are saying, hey, AI is a threat and there are some significant issues to choose to definitely be worried about and think about.

Some of them are a long ways away. So that'd be take takeaway number one from this episode. A lot of myths, some are a long ways away, like the super intelligence thing doesn't mean we ignore it. But we also keep it in perspective. Number two take away, AI can and does provide value to small to medium business owners today. And number three, the scenarios to consider for you to improve are things like improving your sales and your resource usage and your supply chain, things like that. That stuff is very achievable, achievable today. So don't pursue AI in your business. If you can't map it to one of these kinds of scenarios, right, like sales improvement or supply chain and so forth. You need to be able to tie your AI usage, whether that's deep learning or predictive reasoning or image analysis, all that stuff. You need to be able to tie that to some business outcome and if you can't

Then don't do that yet regarding the AI control concerns that I started off with well, and until next time, I wouldn't worry about it, I'd focus on improving your business with AI and don't get too wrapped up with a lot of the AI threat or control hype. Hey, this is Grant thank you for joining looking forward to joining you on the next episode of ClickAI Radio.

Thank you for joining Grant on ClickAI Radio. Don't forget to subscribe and leave feedback. And remember to download your FREE eBook, visit now.

98 つのエピソード