00:00:00
You know 2025 has really seen a surge in the interest and the investment around artificial intelligence and the
00:00:07
expectation is that this technology will continue to grow and lead the way in how
00:00:12
our country and realistically the globe will operate. But with so much investment so quickly there are those
00:00:19
who wonder if we are not headed for a bubble that may see the floor taking right out from under the feet of
00:00:25
artificial intelligence. Not everybody believes that though, or at least that the bubble should be a major concern.
00:00:32
Uh, Lin Woo is an associate professor of operations, information, and decisions here at the Wharton School. Hi, Lynn.
00:00:37
How are you today? >> Good. How are you? Thank you for having me. >> So, in reading some of your comments,
00:00:43
correct me if I'm wrong, you believe there is a bubble, but maybe not a bubble that we should be all totally
00:00:50
concerned about right now. Yes, I believe that is uh so that's that represent my sentiment. It is a bubble.
00:00:58
I'm not going to dispute that. I mean the fact that OpenAI um are doing some
00:01:03
kind of like one to two 1.5 to2 trillion dollar deals uh only having 13 billion dollars in revenue about it's like about
00:01:12
hundred times less than was needed for their investment. So I'm not disputing
00:01:16
as a bubble but I think um it's not a bubble such like the subprime loan where
00:01:22
assets price gets inflated. Uh in this case the bubble is one of these bubble precedes most important technology
00:01:32
evolutions we have in history. So with a technology such as artificial intelligence or genai
00:01:42
that is quickly changing how the thing how our things are being done in every field every industry that needs to be a
00:01:49
lot of investment needed to be for us to get there and that's investment in some
00:01:54
sense that's kind of needed and that's bubble is representation of that so let
00:02:00
me just give you an example by what I mean by this um For example, one of the first technology crash is actually the
00:02:08
railroads, Madak in the 1840s, I believe, right? And uh it was a I believe one maybe one of the first tech
00:02:15
bubbles out there and hundreds of these um maybe even thousands of these uh rail
00:02:20
railway lines that has almost zero economic justification were added anyway, right? And the bubble bursted.
00:02:28
But these railways are really important for Britain for the US became a major transportation network that fueled
00:02:35
future economic growth. Right? So, and same thing with electricity and if you think about a dot uh bubble uh that's in
00:02:44
most recent history that probably the closest one we have today, right? And it's the story is very similar. Do do
00:02:50
then these large kind of transformative technologies when they kind of come on board do they lend themselves at least
00:02:58
to the thought process of a potential bubble just because they're so new just
00:03:03
because they are so different than what we are used to have been experiencing. Well, I think it's almost a necessity.
00:03:10
People have to believe in this medic in order to make that enormous investment, right? So, in do, yeah, lot of company
00:03:18
went belly up after, you know, the 2002, 2003s, right? >> Right. >> But enormous telecom investment, the
00:03:25
fiber optic cables were being built allow that pretty much allowed internet to be so cheap were almost free as we
00:03:32
have today, >> right? And that's eventually as a result of that uh we have e-commerce we had
00:03:39
Google and we have you know you know lots of like we have Netflix right we have lots of these amazing companies
00:03:45
that born right after that because of this economic investment that B and that made on.com
00:03:51
>> well and in in this instance with artificial intelligence we're still not
00:03:54
at a point yet where we have maximized what AI could potentially do and so there's that uncertainty that's
00:04:02
involved. Yeah, absolutely. And let's say in the the worst case, right? So it
00:04:06
bubble bursts, right? But we have made so much investment data center in energy infrastructure that we just haven't been
00:04:13
able to made as a country and even it b even if it bursted one day. Okay. Um you
00:04:19
know and there might be some kind of short-term economic downside for the speculators but the fact is we have
00:04:25
invested a ton of infrastructures that will have tremendous spillover effects on the breast of economy. Right? So if
00:04:32
you just want to invest them on a very small factor of of uh of economy like for example just for housing right it
00:04:40
may or just the asset inflation of housing it didn't really make any kind of infrastructure investment then it
00:04:45
didn't really then then is a probably a bad bubble but in this case a bubble has
00:04:52
made collective investment on the important infrastructure we all need and that positive spillover effects is going
00:04:57
to be you know affecting rest of the economy in a very positive way and >> and you have so many companies that are,
00:05:02
you know, m taking the time and making even a small investment in how AI could have a positive impact on their business
00:05:11
or if it's not going to have that much of a business at least they have researched it and they've looked at it
00:05:17
taking the time and understanding is part of this process. >> Absolutely. And this time is actually
00:05:22
really hard and because the upside is so enormous, >> right? >> But the probability of success is so low
00:05:28
right now, right? there's some recent statistics saying that you know 90 to 90
00:05:33
95% of AI projects fail um depends on how you count it but they but they didn't but they didn't mention perhaps
00:05:40
that the 5% or 10% that were successful are enormously successful they're hundreds of billions of dollar hundreds
00:05:47
of millions of dollars in profit so people want to be the 5% >> how then does it play into this concern
00:05:56
of a bubble that with something like chat GP BT you have updated versions of it seemingly coming out every month, two
00:06:04
months, you know, and and just as a business or a person is getting used to one platform, they have to get used to
00:06:11
something uh different as well. Yeah, that's a part of exciting part about like over like you know maybe
00:06:17
Greenspan said rational exuberance right we are so excited about the technology is going the improve the capability is
00:06:24
improving at an exponential rate that is why you see this uh you know every uh version a new version coming out every
00:06:31
few months because they have to because Gemini is doing it and meta is doing Amazon is trying to do it it's a
00:06:38
collective effort trying to achieve something like AGI that's great whether we're successful or not um is another
00:06:46
story. But during that time, enormous investment for the future is being made and generated. I think we're all going
00:06:51
to benefit from that someday. I >> it it's just unique that we have seen,
00:06:56
as you alluded to before, instances of bubbles that obviously were negative and that we may be in a situation here where
00:07:03
there might be a bubble developing or there is one, but it's not necessarily a
00:07:08
bad thing. Yes, in long run for sure it's probably strictly positive because
00:07:13
the enormous positive spillover from this particular race to do better AI that may made made a bet better energy
00:07:21
access better you know uh data centers that's all going to power future uh you
00:07:26
know economic growth >> does the speed at which this occurs is that factoring into
00:07:32
people's concern about any kind of developing bubble >> well this I mean we rarely seen this
00:07:40
kind of uh speed right I mean we have been experiencing Moore's law of you know doubling every two years or so but
00:07:47
this is like data center uses a 100x right so well I think we're on a some of
00:07:52
a new territory here and uh chat GBT's adoption is like the fastest ever for
00:07:57
the consumer product so I think we're all in a brave new world in the wild west here to see what's actually going
00:08:03
to h what's going to happen but but history you know tell us is not as you know as dire as some of them ne how some
00:08:10
some some have uh you know postured. >> So as this continues to develop are
00:08:15
there things you are watchful for to see play out and to be watching how that bubble kind of develops in the future.
00:08:25
>> Um so I I'm really interested to see positive business use cases like how is
00:08:29
it actually being used? What what are some best business cases in what industry? How they're being developed
00:08:35
the problem associated with it. Once I see this positive cases, I'm my my concern for bubbles are alleviating. But
00:08:42
I do but if I if I see a lot of company just you know um using we haven't quickly figured out how to use these uh
00:08:49
technology in the positive effective way then that's when I get a little worried
00:08:53
but but I have seen some really great positive cases in recent history. Yeah. >> Lynn, great to talk to you again. Thanks
00:08:58
very much for your time today. All the best. >> Yeah, thank you. Lin Woo, associate
00:09:03
professor of operations, information, and decisions here at the Wharton School.