WEBVTT - Using Artificial Intelligence for a Competitive Advantage

0:00:00.080 --> 0:00:03.320
<v Speaker 1>You're listening to Bloomberg Business Week with Carol Messer and

0:00:03.400 --> 0:00:07.200
<v Speaker 1>Tim Stenovic on Bloomberg Radio. We did say, we did

0:00:07.240 --> 0:00:09.000
<v Speaker 1>promise we're going to that we were going to go

0:00:09.080 --> 0:00:11.000
<v Speaker 1>all in on AI. Today we're living up to that.

0:00:11.280 --> 0:00:13.920
<v Speaker 1>Earlier respect to Ian King about how Nvidia is Wall

0:00:13.920 --> 0:00:17.400
<v Speaker 1>Street's top pic for chat gbt mania. After all, Microsoft

0:00:17.840 --> 0:00:21.200
<v Speaker 1>is investing ten billion in Open Ai, the company behind

0:00:21.680 --> 0:00:24.200
<v Speaker 1>the AI tool chat GBT And I do feel like,

0:00:24.720 --> 0:00:28.800
<v Speaker 1>you know, whether you look at major news, you know,

0:00:29.000 --> 0:00:31.479
<v Speaker 1>the networks, the general network networks kind of just the

0:00:31.520 --> 0:00:34.320
<v Speaker 1>general public. Everybody's like, wait a minute, oh AI is

0:00:34.400 --> 0:00:36.959
<v Speaker 1>going on. I didn't realize it was that smart. Yeah,

0:00:37.000 --> 0:00:39.680
<v Speaker 1>they should have checked in with our next guest, because

0:00:39.720 --> 0:00:42.680
<v Speaker 1>then they probably would have known. Ray. Please step with us.

0:00:42.680 --> 0:00:46.920
<v Speaker 1>Tom Davenport, Professor of Information Technology and Management at Babson College.

0:00:47.120 --> 0:00:49.040
<v Speaker 1>He's the co author of the new book All In

0:00:49.120 --> 0:00:52.520
<v Speaker 1>on AI, How Smart Companies Win Big with Artificial Intelligence.

0:00:52.880 --> 0:00:55.840
<v Speaker 1>He joins us via zoom from beautiful Santa Barbara, California.

0:00:55.920 --> 0:00:57.680
<v Speaker 1>Right now, our professor, good to have you with us.

0:00:57.720 --> 0:01:00.720
<v Speaker 1>How are you, thanks, I'm great, happy to be here. Well,

0:01:00.760 --> 0:01:03.280
<v Speaker 1>we're looking forward to going on this deep dive with

0:01:03.320 --> 0:01:05.000
<v Speaker 1>you when it comes to a I I want to

0:01:05.000 --> 0:01:07.320
<v Speaker 1>start with what Carol was referring to, this idea that

0:01:07.360 --> 0:01:10.400
<v Speaker 1>you know, chat gbt GPT has been all over U

0:01:10.640 --> 0:01:13.920
<v Speaker 1>when it comes to sort of like the conversation right

0:01:14.000 --> 0:01:16.920
<v Speaker 1>the you know what we're seeing out in the environment

0:01:17.000 --> 0:01:18.840
<v Speaker 1>right now with with with not just you know, the

0:01:18.840 --> 0:01:20.560
<v Speaker 1>tech side of things and what happened on Twitter a

0:01:20.560 --> 0:01:22.840
<v Speaker 1>couple of months ago when it was released, but this

0:01:22.880 --> 0:01:25.640
<v Speaker 1>idea that hey, this incredible technology is actually out there

0:01:25.640 --> 0:01:27.600
<v Speaker 1>and it's a lot closer to prime time than a

0:01:27.680 --> 0:01:29.839
<v Speaker 1>lot of people thought. What have your thoughts been around

0:01:29.880 --> 0:01:33.880
<v Speaker 1>all this chat GPT? I've never heard of that could

0:01:34.120 --> 0:01:37.559
<v Speaker 1>tell tell any more about it. Really little joke there. Okay,

0:01:39.920 --> 0:01:44.480
<v Speaker 1>you have to be ripped, Ben Winkled, you know what

0:01:45.319 --> 0:01:47.520
<v Speaker 1>I have to say? Though, up until a week or

0:01:47.560 --> 0:01:52.760
<v Speaker 1>so ago, it wasn't necessarily mainstream. I think that's true.

0:01:52.920 --> 0:01:56.200
<v Speaker 1>I wrote an article for Harvard Business Review, I think

0:01:56.240 --> 0:02:01.480
<v Speaker 1>in November on generative technology than what to do about them,

0:02:01.640 --> 0:02:04.280
<v Speaker 1>and CHET GPT came out a couple of weeks later,

0:02:04.320 --> 0:02:07.840
<v Speaker 1>and the whole category just exploded. Um So I kind

0:02:07.840 --> 0:02:10.360
<v Speaker 1>of wish I'd waited a little while, but Yeah, I

0:02:10.400 --> 0:02:12.880
<v Speaker 1>think that it's a very powerful tool, and I think

0:02:12.919 --> 0:02:14.280
<v Speaker 1>it's going to have a lot of impact on a

0:02:14.280 --> 0:02:19.120
<v Speaker 1>lot of sectors of business and society. Um, I think

0:02:19.200 --> 0:02:21.840
<v Speaker 1>we're gonna have to train everybody how to use it,

0:02:21.919 --> 0:02:25.079
<v Speaker 1>or it will be incredibly unfair to those who do

0:02:25.160 --> 0:02:27.120
<v Speaker 1>and those who what do you mean, what do you

0:02:27.120 --> 0:02:33.040
<v Speaker 1>mean by that? Well, it's a huge productivity aid. It'll

0:02:33.080 --> 0:02:36.920
<v Speaker 1>get only better. And so you imagine two kids in

0:02:36.960 --> 0:02:41.520
<v Speaker 1>school and one uses it for preparing essays and the

0:02:41.600 --> 0:02:44.600
<v Speaker 1>other doesn't have access to it. It's just really unfair.

0:02:45.240 --> 0:02:49.160
<v Speaker 1>And you know, I think any business should be exploring

0:02:49.160 --> 0:02:53.280
<v Speaker 1>it aggressively. Now they're just so many different interesting I

0:02:53.280 --> 0:02:55.320
<v Speaker 1>want to make sure I understand this right. It seems

0:02:55.360 --> 0:02:57.520
<v Speaker 1>like you're saying the person who's who's using it to

0:02:57.520 --> 0:02:59.679
<v Speaker 1>help prepare an essay? To me, you know what, I

0:02:59.720 --> 0:03:02.840
<v Speaker 1>think to lot of teachers or school distructor school administrators

0:03:02.800 --> 0:03:06.880
<v Speaker 1>would say that's cheating. Well, where I think we're gonna

0:03:06.960 --> 0:03:14.640
<v Speaker 1>have to revise that traditional expectation. Um, it's a It

0:03:14.720 --> 0:03:18.520
<v Speaker 1>will be virtually impossible to tell. I guess um. The

0:03:18.600 --> 0:03:21.200
<v Speaker 1>only way you could tell is if you know Johnny

0:03:21.240 --> 0:03:24.640
<v Speaker 1>used to write bad essays and now he writes good essays. Um.

0:03:24.800 --> 0:03:32.680
<v Speaker 1>But um, particularly if someone edits the output of chat,

0:03:32.800 --> 0:03:37.160
<v Speaker 1>GPT or whatever generative a tool they're using. Um, you know,

0:03:37.240 --> 0:03:41.040
<v Speaker 1>I think it'll be impossible to tell and really, um,

0:03:41.080 --> 0:03:43.880
<v Speaker 1>not that ultimately different from you know, going to the

0:03:43.920 --> 0:03:46.800
<v Speaker 1>library and reading books. This is the new equivalent of that.

0:03:46.960 --> 0:03:49.440
<v Speaker 1>But does it get Hey, Tom, do we have to

0:03:49.440 --> 0:03:51.960
<v Speaker 1>think about it? It's interesting. I grew up and I

0:03:52.000 --> 0:03:56.480
<v Speaker 1>had to learn penmanship. I remember practicing writing cursive. You know,

0:03:56.920 --> 0:04:01.280
<v Speaker 1>Um that not necessarily was the case for my daughter

0:04:01.560 --> 0:04:04.600
<v Speaker 1>because everything was being done on computers and is being

0:04:04.640 --> 0:04:06.640
<v Speaker 1>done on computers. And I do wonder is there going

0:04:06.680 --> 0:04:10.160
<v Speaker 1>to be kind of well anybody because of something like

0:04:10.200 --> 0:04:13.760
<v Speaker 1>a chat, g g BT or or other large language

0:04:13.840 --> 0:04:17.320
<v Speaker 1>models l m s. You know that it's not going

0:04:17.360 --> 0:04:18.640
<v Speaker 1>to be so important to be able to write an

0:04:18.720 --> 0:04:21.640
<v Speaker 1>essay because a company can have you know, the chat

0:04:21.720 --> 0:04:25.360
<v Speaker 1>gypt do it. So I just wonder, like, how does

0:04:25.440 --> 0:04:33.120
<v Speaker 1>this change transform our world for the better for the worst. Well,

0:04:33.160 --> 0:04:36.039
<v Speaker 1>you know, there are all sorts of possibilities for mischief

0:04:36.240 --> 0:04:40.080
<v Speaker 1>with it. Um. Right now, it's more the image oriented

0:04:40.120 --> 0:04:44.400
<v Speaker 1>tools that are problematic. Although I have heard that hackers

0:04:44.440 --> 0:04:47.960
<v Speaker 1>are already using um GPT three, the sort of a

0:04:48.120 --> 0:04:53.080
<v Speaker 1>parent of chat gpt to create malware UM code, since

0:04:53.120 --> 0:04:55.200
<v Speaker 1>they can write code. But you know, I think in

0:04:55.240 --> 0:04:59.000
<v Speaker 1>general we're all going to become editors rather than creators

0:04:59.000 --> 0:05:04.359
<v Speaker 1>of first draft and um will be UM, we'll have

0:05:04.440 --> 0:05:06.560
<v Speaker 1>to be prompt engineers and you have to write a

0:05:06.600 --> 0:05:09.680
<v Speaker 1>prompt to get these systems to produce something. We'll have

0:05:09.720 --> 0:05:12.919
<v Speaker 1>to be prompt engineers instead of first draft writers. And

0:05:12.920 --> 0:05:15.440
<v Speaker 1>then we will have to edit it to make sure

0:05:15.560 --> 0:05:19.039
<v Speaker 1>that um A, the information is actually correct. Often it

0:05:19.200 --> 0:05:22.359
<v Speaker 1>is not now and um that it's not. You know,

0:05:22.680 --> 0:05:24.880
<v Speaker 1>it can be a little boring at times. You might

0:05:24.920 --> 0:05:27.359
<v Speaker 1>need to liven it up, but in almost every case

0:05:28.160 --> 0:05:31.320
<v Speaker 1>my experience, uh, it comes up with things I hadn't

0:05:31.400 --> 0:05:34.960
<v Speaker 1>thought of, and um, you know, I'm fairly well educated,

0:05:35.120 --> 0:05:38.719
<v Speaker 1>so I think most people would benefit from it. Does

0:05:38.720 --> 0:05:42.559
<v Speaker 1>this scare you at all? Uh? You know, I try

0:05:42.600 --> 0:05:45.400
<v Speaker 1>to look on the positive side. I think there could

0:05:45.440 --> 0:05:50.680
<v Speaker 1>be some really problematic aspects. But this is UM with us. Now.

0:05:50.760 --> 0:05:53.159
<v Speaker 1>We's not going to go away. We might as well

0:05:53.240 --> 0:05:55.279
<v Speaker 1>get used to it and the cats out of the bag,

0:05:55.320 --> 0:05:59.560
<v Speaker 1>you know. Yeah, it's better than having chips implanted in

0:05:59.560 --> 0:06:01.880
<v Speaker 1>our brain, which you know may happen at sometimes. I

0:06:01.960 --> 0:06:03.640
<v Speaker 1>was going to say that's actually you know, if Elon

0:06:03.680 --> 0:06:06.080
<v Speaker 1>Musk has his way, that's that's coming pretty soon thanks

0:06:06.120 --> 0:06:08.800
<v Speaker 1>to you know, what he's working on. Well. It's interesting though,

0:06:08.880 --> 0:06:11.279
<v Speaker 1>you know, there's been a lot of stories about things

0:06:11.320 --> 0:06:14.360
<v Speaker 1>like AI, and I think in terms of the medical community,

0:06:14.360 --> 0:06:17.720
<v Speaker 1>where it can help screen in an emergency room, right

0:06:17.800 --> 0:06:22.080
<v Speaker 1>and help a doctor, um, kind of prioritize or deal

0:06:22.120 --> 0:06:26.040
<v Speaker 1>with more patients in a smart way. In the investment world,

0:06:26.240 --> 0:06:29.560
<v Speaker 1>you know, startup companies that were maybe deemed too small

0:06:29.600 --> 0:06:31.920
<v Speaker 1>and not worthy of spending some time to you know,

0:06:32.360 --> 0:06:34.520
<v Speaker 1>research whether or not that they were even maybe a

0:06:34.560 --> 0:06:37.440
<v Speaker 1>possible good idea that with the use of AI, that

0:06:37.480 --> 0:06:40.120
<v Speaker 1>can be used as a screening force to allow an

0:06:40.120 --> 0:06:43.080
<v Speaker 1>investor to kind of look at more companies. Like I

0:06:43.120 --> 0:06:45.000
<v Speaker 1>think we have to be smart, right with all of

0:06:45.040 --> 0:06:50.039
<v Speaker 1>this stuff. That technology has its pluses and its minuses,

0:06:50.080 --> 0:06:53.200
<v Speaker 1>but there are things that can maybe benefit the world

0:06:53.279 --> 0:06:59.000
<v Speaker 1>more broadly. Oh yeah, a huge potential applications and healthcare,

0:06:59.240 --> 0:07:03.640
<v Speaker 1>although the adoption in clinical practice has been quite slow.

0:07:03.920 --> 0:07:06.800
<v Speaker 1>You know, Um, once a week you get an announcement

0:07:06.800 --> 0:07:09.279
<v Speaker 1>of research that since an AI system is as good

0:07:09.279 --> 0:07:13.840
<v Speaker 1>as are better than a radiologist or I don't know,

0:07:14.040 --> 0:07:19.880
<v Speaker 1>internal medicine specialists at detecting some UM ailment. So um, yeah,

0:07:20.000 --> 0:07:24.600
<v Speaker 1>lots of of potential value. And you know, I think

0:07:24.640 --> 0:07:26.920
<v Speaker 1>it's going to change a lot of jobs. I don't

0:07:26.960 --> 0:07:32.160
<v Speaker 1>think yet it has reduced UM employment much outside of

0:07:32.280 --> 0:07:36.560
<v Speaker 1>you know, maybe robots and manufacturing UM. But you know,

0:07:36.680 --> 0:07:38.680
<v Speaker 1>the people who are going to lose their jobs will

0:07:38.720 --> 0:07:42.520
<v Speaker 1>be the ones who refused to work with AI. Well, Sam,

0:07:42.560 --> 0:07:44.120
<v Speaker 1>you and I were talking about in terms of the

0:07:44.160 --> 0:07:47.480
<v Speaker 1>healthcare community, right, Like you think about all the research

0:07:47.520 --> 0:07:52.440
<v Speaker 1>that gets constantly you know, done and results and you

0:07:52.480 --> 0:07:55.160
<v Speaker 1>know papers and stuff like how can a doctor who's

0:07:55.200 --> 0:07:57.440
<v Speaker 1>taking care of patients possibly read it all? So you

0:07:57.440 --> 0:07:59.720
<v Speaker 1>think about if AI can be used as a tool

0:08:00.160 --> 0:08:02.000
<v Speaker 1>to help screen through some of this, I mean think

0:08:02.040 --> 0:08:04.160
<v Speaker 1>about the benefits of it. Yeah, I mean I think

0:08:04.160 --> 0:08:06.520
<v Speaker 1>that's really important case. But I also think, you know,

0:08:06.560 --> 0:08:09.640
<v Speaker 1>one of the problems is is you know what Professor

0:08:09.680 --> 0:08:12.679
<v Speaker 1>Davenport talked about, and it's the idea that okay, well

0:08:13.000 --> 0:08:16.560
<v Speaker 1>you can use chat GPT too. Also you know, create

0:08:16.800 --> 0:08:19.800
<v Speaker 1>malware or or write malicious code. There's you know, there's

0:08:19.800 --> 0:08:22.960
<v Speaker 1>sort of two sides to every coin, professor, So do

0:08:23.000 --> 0:08:25.040
<v Speaker 1>you think there needs to be any sort of regulation

0:08:25.080 --> 0:08:32.160
<v Speaker 1>in this type of technology? I think that for the momentum,

0:08:32.240 --> 0:08:35.480
<v Speaker 1>individual organizations will have to decide, you know, what's our

0:08:35.480 --> 0:08:39.920
<v Speaker 1>policy on the schools. As we were discussing earlier. UM,

0:08:40.120 --> 0:08:45.040
<v Speaker 1>I don't see the US government regulating effectively anytime soon.

0:08:45.160 --> 0:08:48.520
<v Speaker 1>You know, possibly we can see something in Europe, although

0:08:48.520 --> 0:08:52.480
<v Speaker 1>I think this is going to be really difficult to regulate, um,

0:08:52.520 --> 0:08:56.960
<v Speaker 1>even for you know, highly regulatory societies. And I think

0:08:57.640 --> 0:09:01.840
<v Speaker 1>we've already seen apparently some student who used to work

0:09:01.880 --> 0:09:04.800
<v Speaker 1>for open ai has come up with a system that

0:09:04.920 --> 0:09:08.520
<v Speaker 1>the text whether or not a passage was written by

0:09:08.640 --> 0:09:12.040
<v Speaker 1>chat GPT. So um, you know, maybe we'll have a

0:09:12.120 --> 0:09:15.200
<v Speaker 1>sort of ongoing arms race in that regard. Hey, we're

0:09:15.200 --> 0:09:17.920
<v Speaker 1>going to come back with Tom Davenport, professor of Information

0:09:17.960 --> 0:09:20.640
<v Speaker 1>Technology Management at babs In College. His book. He's co

0:09:20.720 --> 0:09:24.079
<v Speaker 1>author on the book entitled All In on AI, How

0:09:24.120 --> 0:09:28.040
<v Speaker 1>Smart Companies Win Big with Artificial Intelligence. And in the book,

0:09:28.200 --> 0:09:30.680
<v Speaker 1>you know, gets into some cases, whether it's a Disney,

0:09:30.679 --> 0:09:33.400
<v Speaker 1>a Walmart, a Capital one of Fiser and Eli Lily,

0:09:33.400 --> 0:09:36.080
<v Speaker 1>how they've been using AI. So we'll talk about that.

0:09:36.200 --> 0:09:39.640
<v Speaker 1>On the other side, Carol Master along with Tim Stanivic

0:09:39.760 --> 0:09:43.120
<v Speaker 1>live in our Bloomberg Interactive Broker studio. Tim just reminding me,

0:09:43.240 --> 0:09:46.120
<v Speaker 1>I've been saying chat GPT, Where the hell did it

0:09:46.240 --> 0:09:50.439
<v Speaker 1>come from? GPT? Potato potato, Carolo tomato coming out of

0:09:50.480 --> 0:09:52.959
<v Speaker 1>my brain. Maybe if you had more experience with AI,

0:09:53.040 --> 0:09:54.880
<v Speaker 1>you wouldn't have made that mistake. Maybe if a I

0:09:54.960 --> 0:09:57.640
<v Speaker 1>was running me, I wouldn't. That's definitely true. All right,

0:09:57.720 --> 0:09:59.880
<v Speaker 1>let's get back to our guest. Tom Davenport, Professor in

0:10:00.040 --> 0:10:03.720
<v Speaker 1>Ormation Technology Management at Babson College, co author of the

0:10:03.720 --> 0:10:07.120
<v Speaker 1>book All In on AI, How Smart Companies Win Big

0:10:07.160 --> 0:10:11.120
<v Speaker 1>with Artificial Intelligence, Still with us via zoom from Santa Barbara, California.

0:10:11.400 --> 0:10:13.080
<v Speaker 1>You know, tell one thing I do want to go

0:10:13.120 --> 0:10:15.120
<v Speaker 1>back to, and it's funny. Um. In the break, I

0:10:15.160 --> 0:10:19.360
<v Speaker 1>went and actually was doing some additional googling and searching.

0:10:19.600 --> 0:10:23.920
<v Speaker 1>Came across the story by Ardina Bass and a colleague

0:10:24.240 --> 0:10:27.360
<v Speaker 1>on kind of what is artificial intelligence? Not so long

0:10:27.400 --> 0:10:30.360
<v Speaker 1>ago when we think about artificial intelligence and sounds so

0:10:30.360 --> 0:10:35.360
<v Speaker 1>sci fi if you will, but it's really not, is it. No,

0:10:35.760 --> 0:10:38.560
<v Speaker 1>It's in you know a lot of devices that we

0:10:38.720 --> 0:10:41.360
<v Speaker 1>use every day. Um. You know, it's I don't think

0:10:41.400 --> 0:10:45.280
<v Speaker 1>there's anything terribly mysterious about it. It's just using technology

0:10:45.320 --> 0:10:47.960
<v Speaker 1>to do things that were ordinarily just done by the

0:10:48.040 --> 0:10:50.840
<v Speaker 1>human brain. And I suppose arguably we've been doing that

0:10:50.880 --> 0:10:55.040
<v Speaker 1>since you know, calculators did long division for us. Who's

0:10:55.080 --> 0:10:57.680
<v Speaker 1>doing this really well, you know, with a lot of

0:10:57.679 --> 0:10:59.719
<v Speaker 1>the attention is kind of Microsoft because of its huge

0:10:59.720 --> 0:11:03.440
<v Speaker 1>and vest smith in open ai and chat GPT. But

0:11:04.160 --> 0:11:08.840
<v Speaker 1>what companies are deploying this technology really well? Well, we

0:11:09.080 --> 0:11:11.920
<v Speaker 1>focused in the book on not on the digital native

0:11:12.000 --> 0:11:14.880
<v Speaker 1>companies because you know, it's relatively easy for them. They've

0:11:14.880 --> 0:11:16.400
<v Speaker 1>been doing it for a while. They didn't have a

0:11:16.640 --> 0:11:20.520
<v Speaker 1>kind of a technical debt, uh, a base of previous

0:11:20.600 --> 0:11:22.600
<v Speaker 1>technologies that they had to deal with. They didn't have

0:11:22.679 --> 0:11:27.360
<v Speaker 1>to UM persuade people that it was important. So UM

0:11:27.480 --> 0:11:32.080
<v Speaker 1>in we found a variety of companies in different industries

0:11:32.320 --> 0:11:42.040
<v Speaker 1>in UM consumer products and retail. There's Unilever um uh uh.

0:11:42.880 --> 0:11:46.760
<v Speaker 1>Kroger quite good at this is this Are they using

0:11:46.760 --> 0:11:50.000
<v Speaker 1>this to help identify what demand is going to be

0:11:50.520 --> 0:11:53.199
<v Speaker 1>in a certain area of the country based on and

0:11:53.800 --> 0:11:58.439
<v Speaker 1>every area, Yeah, and every area of the country. Basically,

0:11:58.559 --> 0:12:03.360
<v Speaker 1>Kroger does a prediction of every skew, every stock keeping

0:12:03.520 --> 0:12:09.040
<v Speaker 1>unit in every store every night to make sure they

0:12:09.040 --> 0:12:11.400
<v Speaker 1>don't have stockouts, and you know that they get the

0:12:11.400 --> 0:12:15.960
<v Speaker 1>product to the store. So that's just too much data

0:12:16.040 --> 0:12:18.760
<v Speaker 1>to ever do well with a human brain. So you

0:12:18.800 --> 0:12:21.120
<v Speaker 1>have to use machine learning to to make those kinds

0:12:21.120 --> 0:12:24.200
<v Speaker 1>of predictions, and they're quite successful at that. They do

0:12:24.840 --> 0:12:29.720
<v Speaker 1>UM over a billion loyalty offers a year based on

0:12:29.960 --> 0:12:31.679
<v Speaker 1>you know, what they think you might want to want

0:12:31.720 --> 0:12:34.000
<v Speaker 1>to buy, if you remember their loyalty program, based on

0:12:34.040 --> 0:12:36.600
<v Speaker 1>what you bought in the past, So they're really good

0:12:36.640 --> 0:12:40.520
<v Speaker 1>at it. Of course, some banks Capital One historically that

0:12:40.720 --> 0:12:44.040
<v Speaker 1>wast analytical bank, and and I think probably the best

0:12:44.080 --> 0:12:49.199
<v Speaker 1>certainly for its size, at Ai Shell air Bus, UM,

0:12:49.280 --> 0:12:53.280
<v Speaker 1>Elevant's Health which used to be Anthem is a health

0:12:53.280 --> 0:12:57.320
<v Speaker 1>insurance company that's really focused on this UM and then

0:12:57.600 --> 0:13:02.240
<v Speaker 1>UM in Canada. Lob Laws are just retailer in Canada.

0:13:02.400 --> 0:13:04.839
<v Speaker 1>UM a couple of banks we found Coacher Bank is

0:13:04.920 --> 0:13:11.599
<v Speaker 1>quite good, UM, DBS Bank in Singapore really fantastic. And yeah,

0:13:11.640 --> 0:13:14.880
<v Speaker 1>I mean less than one per cent of companies, i'd say,

0:13:14.920 --> 0:13:16.960
<v Speaker 1>but good number. But a lot of the big ones

0:13:17.000 --> 0:13:19.160
<v Speaker 1>we've heard of well and what's interesting is you shared

0:13:19.200 --> 0:13:22.160
<v Speaker 1>with our producer Paul. You know. Um, over the last

0:13:22.200 --> 0:13:26.200
<v Speaker 1>five years, companies integrating AI extensively throughout their organizations have

0:13:26.280 --> 0:13:29.320
<v Speaker 1>experienced stock prices averaging four times the performance of the

0:13:29.440 --> 0:13:32.240
<v Speaker 1>S and P five hundred. Yet this AI field group

0:13:32.280 --> 0:13:34.440
<v Speaker 1>accounts for less than one percent of large companies, so

0:13:34.600 --> 0:13:37.520
<v Speaker 1>little perspective. Where's this research? Who did this research in

0:13:37.600 --> 0:13:42.160
<v Speaker 1>terms of the share prices? That was done by Deloitte? Um,

0:13:42.559 --> 0:13:45.360
<v Speaker 1>my co author is the head of AI for Deloitte.

0:13:46.120 --> 0:13:49.840
<v Speaker 1>So why do you think that's happening? Well, you know,

0:13:50.640 --> 0:13:55.559
<v Speaker 1>AI pays off. It means that you have better pricing

0:13:56.320 --> 0:13:59.920
<v Speaker 1>of your products, better offers. Increasingly, AI is and bed

0:14:00.040 --> 0:14:05.280
<v Speaker 1>it into products and services. Um. We talk about Morgan

0:14:05.400 --> 0:14:10.240
<v Speaker 1>Stanley using it for quote next best Action for their

0:14:10.320 --> 0:14:15.040
<v Speaker 1>financial advisors to their to their clients. Um Uh. Some

0:14:15.120 --> 0:14:18.280
<v Speaker 1>companies I think probably couldn't even do what they do

0:14:18.360 --> 0:14:20.880
<v Speaker 1>at all. We talk about a small to medium sized

0:14:20.920 --> 0:14:25.240
<v Speaker 1>company called c c C Intelligent Solutions that can give

0:14:25.320 --> 0:14:29.160
<v Speaker 1>you through companies like USA, can give you an estimate

0:14:29.240 --> 0:14:32.720
<v Speaker 1>of what you're crashed car will cost a fix in

0:14:32.840 --> 0:14:36.920
<v Speaker 1>less than a minute based on image recognition and AI

0:14:37.080 --> 0:14:40.320
<v Speaker 1>and a vast amount of data. I also feel like,

0:14:40.360 --> 0:14:42.720
<v Speaker 1>you know, when we talk about anything data related, Tom

0:14:43.120 --> 0:14:45.120
<v Speaker 1>is that the data is only as good as the input, right.

0:14:45.160 --> 0:14:47.720
<v Speaker 1>And we've talked a lot about biased nous or bias

0:14:47.840 --> 0:14:50.240
<v Speaker 1>excuse me, in terms of in data. And we talked

0:14:50.280 --> 0:14:53.560
<v Speaker 1>about this a lot coming off of the death of

0:14:53.560 --> 0:14:56.760
<v Speaker 1>George Floyd and lack of diversity within our world. I mean,

0:14:56.760 --> 0:15:00.160
<v Speaker 1>it's been a continued story for many, many years. But

0:15:00.200 --> 0:15:02.120
<v Speaker 1>now the data is involved. You know, you do worry

0:15:02.120 --> 0:15:05.400
<v Speaker 1>about how things are skewed. How do we ensure that

0:15:05.480 --> 0:15:11.880
<v Speaker 1>AI is pure? Well, the best companies now, you know, UM,

0:15:11.960 --> 0:15:15.440
<v Speaker 1>you have to move beyond the kind of proclamations that

0:15:15.520 --> 0:15:19.080
<v Speaker 1>we we're going to be you know, responsible with our

0:15:19.120 --> 0:15:23.160
<v Speaker 1>AI and a few policy statements we talked in the book,

0:15:23.200 --> 0:15:29.280
<v Speaker 1>for example, about Unilever, which reviews every AI application before

0:15:29.320 --> 0:15:33.600
<v Speaker 1>it's developed to make sure that it's transparent and fair

0:15:33.920 --> 0:15:37.360
<v Speaker 1>not biased. UM. They have a they have a policy

0:15:37.400 --> 0:15:41.080
<v Speaker 1>that you can't really make any final decision for UM

0:15:41.120 --> 0:15:45.640
<v Speaker 1>a customer, employee, or partner without a human being involved,

0:15:45.720 --> 0:15:48.040
<v Speaker 1>and they make sure that that happens before they build

0:15:48.040 --> 0:15:52.760
<v Speaker 1>the application. M does that work, UM? It has? It

0:15:52.800 --> 0:15:56.320
<v Speaker 1>turns out you know, UM. We keep hearing the same

0:15:56.440 --> 0:16:02.320
<v Speaker 1>bad stories of highly biased UM programs like the Amazon

0:16:02.440 --> 0:16:05.040
<v Speaker 1>one that had a preference for male engineers, where with

0:16:05.120 --> 0:16:07.440
<v Speaker 1>them over and over again, UNI Leaver has only had

0:16:07.520 --> 0:16:11.920
<v Speaker 1>two out of over I think two hundred that really

0:16:12.040 --> 0:16:14.400
<v Speaker 1>violated their policies and they had to go back and

0:16:14.800 --> 0:16:17.440
<v Speaker 1>redo them. All right, we're gonna leave it on that. Hey, listen, Tom,

0:16:17.440 --> 0:16:19.440
<v Speaker 1>thanks for sticking around. We were looking to do a

0:16:19.480 --> 0:16:22.080
<v Speaker 1>deep dive into it since it feels like everybody's talking

0:16:22.600 --> 0:16:26.280
<v Speaker 1>about chat GPT now that I remember how to say

0:16:26.320 --> 0:16:28.320
<v Speaker 1>it UM, but it does feel like all of a

0:16:28.320 --> 0:16:30.960
<v Speaker 1>sudden AI it's getting a lot of attention, and it

0:16:31.000 --> 0:16:33.080
<v Speaker 1>helps to really kind of understand exactly what's going on.

0:16:33.160 --> 0:16:37.160
<v Speaker 1>Tom Davenport, Professor of Information Technology and Management at Babson College,

0:16:37.200 --> 0:16:39.880
<v Speaker 1>his new book All In on AI. He's co author

0:16:40.240 --> 0:16:42.360
<v Speaker 1>All In on AI, How Smart Companies Been Big with

0:16:42.480 --> 0:16:46.320
<v Speaker 1>Artificial Intelligence? Joining us via zoom from Santa Barbara, California.

0:16:46.360 --> 0:16:48.560
<v Speaker 1>What are you thinking? I'm thinking this is going to

0:16:48.640 --> 0:16:51.920
<v Speaker 1>be a serious challenge for parents, for teachers. I still

0:16:51.920 --> 0:16:53.240
<v Speaker 1>think about the academics of it.