我们关注错了 AI 竞赛：Oliver Wyman 首席执行官谈适应与控制
We’re Watching the Wrong AI Race
Oliver Wyman Forum 管理合伙人兼 CEO John Romeo 在 2026 年 Bloomberg New Economy Forum 上指出，人们过度关注前沿实验室投入数百亿美元于芯片、模型和数据中心的"智能竞赛"，却忽视了两场同样关键的竞赛：适应与控制。他认为 AI 正从提供答案转向被赋予权威，而随着 AI 更强、更易获取，优势正从模型本身转向数据、工作流和运营。
Bloomberg the Company & Its Products The Company & its ProductsBloomberg Terminal Demo RequestBloomberg Anywhere Remote Login Bloomberg Anywhere LoginBloomberg Customer Support Customer Support
Bloomberg
Connecting decision makers to a dynamic network of information, people and ideas, Bloomberg quickly and accurately delivers business and financial information, news and insight around the world
For Customers
* [Bloomberg Anywhere Remote Login](https://bba.bloomberg.net/?utm_source=bloomberg-menu&utm_medium=bcom)
* [Software Updates](https://www.bloomberg.com/professional/support/customer-support/software-updates/?utm_source=bloomberg-menu&utm_medium=bcom)
* [Manage Products and Account Information](https://service.bloomberg.com/portal/sessions/new?utm_source=bloomberg-menu&utm_medium=bcom)
Support
Americas+1 212 318 2000
EMEA+44 20 7330 7500
Asia Pacific+65 6212 1000
Company
Communications
* [Press Announcements](https://www.bloomberg.com/company/announcements/?utm_source=bloomberg-menu&utm_medium=bcom)
* [Press Contacts](https://www.bloomberg.com/company/press-contacts/?utm_source=bloomberg-menu&utm_medium=bcom)
Follow
* [Facebook](https://www.facebook.com/Bloomberglp)
* [Instagram](https://www.instagram.com/bloomberg/)
* [LinkedIn](https://www.linkedin.com/company/2494)
* [YouTube](https://www.youtube.com/@bloomberglp)
Industry Products
* [Bloomberg Law](https://pro.bloomberglaw.com/?utm_source=bloomberg-menu&utm_medium=bcom)
* [Bloomberg Tax](https://pro.bloombergtax.com/?utm_source=bloomberg-menu&utm_medium=bcom)
* [Bloomberg Government](https://about.bgov.com/?utm_source=bloomberg-menu&utm_medium=bcom)
* [BloombergNEF](https://about.bnef.com/?utm_source=bloomberg-menu&utm_medium=bcom)
Media
Media Services
* [Bloomberg Media Distribution](https://www.bloomberg.com/distribution?utm_source=bloomberg-menu&utm_medium=bcom)
* [Advertising](https://www.bloombergmedia.com/?utm_source=bloomberg-menu&utm_medium=bcom)
Company
Communications
* [Press Announcements](https://www.bloomberg.com/company/announcements/?utm_source=bloomberg-menu&utm_medium=bcom)
* [Press Contacts](https://www.bloomberg.com/company/press-contacts/?utm_source=bloomberg-menu&utm_medium=bcom)
Follow
* [Facebook](https://www.facebook.com/Bloomberglp)
* [Instagram](https://www.instagram.com/bloomberg/)
* [LinkedIn](https://www.linkedin.com/company/2494)
* [YouTube](https://www.youtube.com/@bloomberglp)
Industry Products
* [Bloomberg Law](https://pro.bloomberglaw.com/?utm_source=bloomberg-menu&utm_medium=bcom)
* [Bloomberg Tax](https://pro.bloombergtax.com/?utm_source=bloomberg-menu&utm_medium=bcom)
* [Bloomberg Government](https://about.bgov.com/?utm_source=bloomberg-menu&utm_medium=bcom)
* [BloombergNEF](https://about.bnef.com/?utm_source=bloomberg-menu&utm_medium=bcom)
Media
Media Services
* [Bloomberg Media Distribution](https://www.bloomberg.com/distribution?utm_source=bloomberg-menu&utm_medium=bcom)
* [Advertising](https://www.bloombergmedia.com/?utm_source=bloomberg-menu&utm_medium=bcom)
Bloomberg
Connecting decision makers to a dynamic network of information, people and ideas, Bloomberg quickly and accurately delivers business and financial information, news and insight around the world
For Customers
* [Bloomberg Anywhere Remote Login](https://bba.bloomberg.net/?utm_source=bloomberg-menu&utm_medium=bcom)
* [Software Updates](https://www.bloomberg.com/professional/support/customer-support/software-updates/?utm_source=bloomberg-menu&utm_medium=bcom)
* [Manage Contracts and Orders](https://service.bloomberg.com/portal/sessions/new?utm_source=bloomberg-menu&utm_medium=bcom)
Support
Americas+1 212 318 2000
EMEA+44 20 7330 7500
Asia Pacific+65 6212 1000
Sign InSubscribe
US Edition
Video Player is loading.
Unmute
Current Time 0:02
Loaded: 9.30%
0:02
1x
Playback Rate
2x
1.75x
1.5x
1.25x
1x, selected
0.75x
0.5x
0.25x
captions off, selected
English
This is a modal window.
Ad 0:29
Unmute
Oct 13th, 2026
We’re Watching the Wrong AI Race
John Romeo, Managing Partner & CEO, Oliver Wyman Forum examines how humans can adapt and retain agency as AI becomes more powerful at the 2026 Bloomberg New Economy Forum in New Delhi.
Playlist
We’re Watching the Wrong AI Race Now Playing
Rising US Bond Yields Show No Sign of Slowing Duration:3:43
Defiance CEO's Son Inspires Autism ETF Duration:2:44
US Mortgage Rates Rise for Seven Straight Weeks Duration:4:19
Defiance ETF Looks to Make Impact Against Autism Duration:8:01
10-Year Yield at 6% A Major Psychological Level for Investors: Singaravelu Duration:6:01
Muse Wonder Puts Meta's Aggressive AI Push in Focus Duration:3:27
Nobel Prize in Economics Shared by Hall, Woodford Duration:0:55
Why the Houthis Are So Hard to Defeat Duration:7:12
Neostellar’s Willy Lee on AI Landscape Ahead of Anthropic Pre-IPO Investor Day Duration:6:12
Cisco, Nvidia Tackle AI’s Growing Data Center Bottleneck Duration:7:09
How New NHL, MLB Indexes Might Turn Into ETFs Duration:3:48
Factors Driving Thematic Investing Story: Global X CEO Duration:6:53
How the OpenAI, Anthropic IPOs Could Lift the Veil on AI Economics Duration:5:25
Tesla Cybercab to Make European Debut at Paris Motor Show Duration:3:56
Stellantis CEO Says Company Split 'Not Plausible' Duration:1:32
0:00
Thank you. That the first hint came in 2016 when
0:06
Lee Sedol, the renowned go champion, faced off against Google DeepMind's
0:10
AlphaGo. Now, go had been played for thousands of
0:14
years before this, but the history of the game changed in an instant when
0:19
AlphaGo played move 37. The commentators at first thought it was
0:24
a mistake. No professional player would have dared
0:26
make that that move, but at the end it turned out to be a stroke of brilliance.
0:30
It helped AlphaGo win the game and overturned centuries of conventional
0:34
wisdom. I think a lot about move 37 today,
0:39
because I is often framed as just the next chapter in a familiar technology
0:44
story, going from the computer to the internet to the smartphone.
0:48
And now I, I started by giving us answers, but now we're beginning to give
0:54
it authority. And so something more fundamental is
0:56
happening, and I'm not certain that our companies or countries or regulators are
1:02
ready for that. Now, today, everyone is watching one
1:07
race. You're watching the race for
1:09
intelligence. You have the frontier labs.
1:10
You have us in China of hundreds of billions of dollars getting poured into
1:14
chips, into models, into data centers. But I think we're paying too much
1:19
attention to that race, because there are two other races that I think could
1:23
be just as consequential. The race for adaptation and the race for
1:27
control. Consider adaptation right as I becomes
1:31
more powerful and more accessible. The advantage shifts from the model to
1:36
everything that surrounds the model your data, your workflows, your operations,
1:40
your processes. And now we're seeing it right now.
1:44
We're seeing pilots everywhere. Everyone's experimenting with with AI.
1:48
You see it all across the organization, except perhaps in the bottom line where
1:51
it remains scarce. And it's because I create time.
1:55
It creates capacity, but it doesn't automatically create value.
2:00
The companies that are realizing that value are actually reimagining their
2:03
workflows. More often than not, though, it's just
2:06
pushing the bottleneck from one place to another.
2:13
So Hang on a second. Our new Oliver Wyman Forum research
2:16
shows that only 8% of organizations actually have a standard for what
2:20
happens to that, I save time. An insurance claim can be processed in
2:24
seconds. But if that insurance claim then sits in
2:27
a queue for three days waiting for approval, you've just made one step
2:30
spectacularly efficient. You haven't transformed a company.
2:33
The faster the machine gets, the more visible the rest of the organization
2:38
becomes. Because a million automated processes
2:41
don't create a transformed company. The companies that are creating value
2:45
are changing how they operate. And as I becomes much more capable, the
2:50
question shifts from what can I do to what are we prepared to let into?
2:55
Give an agent a your data, a budget, a mandate, and permission to act.
3:00
And you've created an economic actor. When the travel agent that chooses an
3:04
airline is no longer a human, then the battle that we fought for, human
3:07
attention shifts to a battle for machine permission, and power follows
3:11
permission. And that brings me to the third race,
3:14
the race for control. Tim Berners-Lee made a suggestion to me
3:18
recently a wonderfully pragmatic suggestion.
3:22
He said give agents license plates. So if an agent does harm, we know whose
3:27
it was, who's accountable, what it was allowed to do and who can stop it.
3:32
Identity. Permission.
3:33
Accountability. Revoke ability.
3:35
It's the basis of that. Yuval Noah Harari made a similar
3:38
observation recently. He said people lie and they cheat.
3:41
It's why we created courts. It's why we created contracts.
3:44
Auditors, regulators. Civilization works not because we trust
3:48
everyone, but because we've learned to cooperate with those that we can't
3:51
necessarily completely trust. And the same is true for AI.
3:55
We need institutions that don't focus on making machines that we can blindly
3:59
trust. But the institutions focus on allowing
4:02
us to work with machines that explicitly we don't trust.
4:06
The paradox is, increasingly we're going to need machines that supervise
4:08
machines. It doesn't mean the humans aren't there.
4:11
The humans have to go up a level, because the better we get at setting the
4:14
boundaries and ensuring that revoke ability, then the more confidence we can
4:19
have to delegate to the machines. So it's a machine speed oversight, but
4:24
it's ultimately a human accountability. And it's the same challenge for
4:27
countries. In India, our research shows huge demand
4:30
for building AI is now the national capability on AI, but equally demands to
4:36
ensure access to the world's best models.
4:38
Now, it's not a contradiction. It's a new definition, I think, of
4:41
sovereignty that focuses on emphasizing the permission and preserving choice
4:46
rather than eliminating dependence. We don't need every country to regulate
4:50
AI identically. We just need interoperable
4:52
accountability. But there's a danger in everything that
4:56
I just said, that it makes us sound like spectators, like the only choice that we
5:00
I just said, that it makes us sound like spectators, like the only choice that we
5:00
really have is how quickly we adapt. And many, many employees feel that way.
5:06
68% of employees say they take a pay cut for a guarantee that I would never
5:10
replace them, but we're not spectators.
5:14
Technology has a direction because people give it one.
5:17
Companies decide what machines can do. The government set the rules, but
5:20
ultimately society has to have a voice in what this extraordinary technology is
5:24
for. We know that humans are better at
5:27
developing power than we are necessarily developing the wisdom to use it.
5:31
The Prime Minister talked about that this morning.
5:33
And so this is an important design moment.
5:36
Which brings me back to move 37 and go. We celebrated a machine making a move
5:42
that humans had never conceived of. Right.
5:45
We want I to do that. We want to celebrate that.
5:47
We want AI to discover new medicines, to create new materials, to find solutions
5:52
to problems we can't solve ourselves. But what happens when move 37 comes to a
5:57
financial market, to a payment system, to a military decision?
6:02
The machines may be better at playing the game than we are, but it doesn't
6:06
mean that we have to step away from that.
6:08
We have a responsibility to to write the rules for the games, and indeed to
6:12
determine which games the machines are even allowed to play.
6:16
AI is going to have agency, but ultimately, so are we.
6:20
And the future of AI is not something that we should just sit back and let
6:23
happen to us. We can't be passive.
6:26
This is sort of a moment to to take that control, because we have a
6:29
responsibility to choose that. Thank you.
6:35
For.
Top Video
Duration:1:58 ### Blackstone's Gray Says AI Infrastructure Is 'Right Place to Be'
Duration:3:27 ### Muse Wonder Puts Meta's Aggressive AI Push in Focus
Duration:1:32 ### Trump's Diesel Deal With Putin Fuels New Tension With Ukraine
Duration:4:51 ### IMF Says Growth Should Pick Up With AI Investment Boost
Duration:2:47 ### Hobbs Sees Europe as a Possible Haven as Chips Slump
Duration:1:28 ### AI’s Giant Money Loop Has Echoes of 2008
Duration:3:01 ### What to Expect From Big US Bank Earnings
Duration:1:55 ### Apple to Unveil Its Smart Home Strategy #### Technology
Available on:
Listen on Apple TVListen on RokuListen on Samsung TVListen on Fire TVListen on Android TVListen on Rakuten TVListen on Haystack News
Watch BTV in your area:
Terms of ServiceDo Not Sell or Share My Personal InformationTrademarksPrivacy Policy
Help©2026 Bloomberg L.P. All Rights Reserved.
We've updated our terms
By accepting, you agree to our updated Terms of Service, including the arbitration provision and class action waiver. You understand that we process your information as described in the Privacy Policy, which may include sharing information about your use of Bloomberg.com with third parties. Accepting does not affect your right to opt-out at any time via the Do Not Sell or Share My Personal Information link in the website footer.
Accept
来源：Bloomberg：Technology · bloomberg.com