00:00I am curious about whether some of the warnings that we got, and it wasn't just Dario at Anthropic,
00:05we had a team of researchers at Microsoft today put out a 15,000 word manifesto,
00:09and there have been a few other warnings from some of these CEOs to kind of just slow things down
00:13a little bit.
00:14I'm wondering how that feeds into the sentiment right now amongst the users,
00:19meaning the CEOs that have been deploying all of this stuff internally for their own employee use.
00:24Yeah, and look, I think that's a great question, and the reality is that the debate that's been playing out
00:29over the last sort of weeks, days, et cetera, is a really important one, right?
00:32Where AGI is going to go, what the threats are, where we need pace, yes, but where we also need
00:38patience.
00:39But what's interesting is when you transport yourself into the boardrooms, the management teams and so on,
00:44their question is a different one.
00:45Their question is, OK, I get all of that that's happening on the frontier and we're going to get there,
00:49but right now I've spent a ton of money and I've really got to make sure that everything that I'm
00:55doing
00:55is actually going to translate through to scalability, to profitability, and to sustainability going forward.
01:01So the conversation in the boardroom is not a lot of what we're seeing play out.
01:05It is much more acute than that, and to make sure that we're really getting the bottom line impact from
01:10AI.
01:10Well, talk about that bottom line impact, because, I mean, just looking at the report,
01:1380% of the workers say AI made them more productive.
01:18I assume that's good personally.
01:20But then the share of companies seeing any bottom line impact was basically at 37%,
01:25which is basically flat for the year.
01:28So give me a sense here of why there's that gap between me saying I'm more productive,
01:32but the people upstairs who have to cut the checks are saying, wait a minute, you know, where's the return?
01:38Yeah, and I think a lot of this comes down to the fact that just because you or I
01:42can write our emails faster or could be more productive in our day
01:45does not translate to, you know, companies actually getting more value from this.
01:50And when you step back from it and say, well, you know, why is that happening?
01:52What's the disconnect?
01:53One of the biggest factors is people are still treating this as a technology transformation.
01:58You know, I'm going to pick the right models.
02:00I'm going to, you know, deploy the right technology and good things will flow.
02:03And actually what we find is, you know, an AI transformation is at its heart a people transformation.
02:08It's actually the workflow change, right?
02:11The upskilling and reskilling, the changes in the jobs to be done that makes a far,
02:16far bigger difference than, you know, did I just pick the right technology?
02:19So then what happens to ROI?
02:20Because you do point to the fact that when AI companies pitch, they say, okay, cost will fall.
02:25But you say that's not true.
02:26So if human oversight is still there, how are you going to think about profitability when
02:31we need people and these expensive AI models?
02:34Yeah, that's a great point.
02:35So, you know, the way that I would think about it is the ROI is going to come from actually
02:41reimagining how the work gets done, right?
02:43So what are the jobs to be done?
02:44And then how does that workflow need to change?
02:47So I often say with companies for, you know, for every dollar that you're going to spend
02:50on the technology, there should be an equal and opposite dollar, sometimes many times that,
02:55in actually getting to, well, what's the workflow that now needs to change?
02:59Can I reimagine that, not just re-engineer it?
03:01You know, how do I need to invest in the human wear, if you like, not just the software and
03:05the hardware to make sure that the ROI translates down?
03:09That's your point.
03:09If that, you know, ends up being kind of all additive, then it's a problem.
03:13And this is why the difference between re-engineering a process, you know, looking at can I make
03:17every bit incrementally better and put technology in it, that is different to can I reimagine it?
03:23Are there 25 steps and 15 of them don't actually need to happen anymore?
03:27And you look at it as a hybrid workforce, right?
03:29Right. Agents plus humans together, not trying to tack agents onto the outside of an already
03:35human workflow.
03:36Over the weekend, we've heard from a drumbeat of four innings, Romain detailed them at the
03:39top of the show.
03:40So we have those people in unison, the biggest AI leaders, but we also have some pushback,
03:45so to speak, like from David Sachs.
03:47He said, OK, regulate it.
03:49I mean, no one's stopping you.
03:50And then we also have some prominent people saying it's a little too late.
03:53Where do you sit in the spectrum?
03:54Yeah, I mean, the thing I would say on this is the rate and pace of technology change right
04:00now is outstripping, you know, organizational, regulatory ability to absorb that change.
04:06And that's the piece that I think we've heard over the weekend and past weeks that we need
04:10to start closing the gap of.
04:12And so for me, all of this comes down to where we start at the top of this of there
04:16are some
04:17places where actually pace is really, really needed, right, for maybe not the frontier,
04:21frontier sort of side of things, but regular model use and so on in companies.
04:25Yeah, we need to speed up the pace of translating the 80% to the 37%.
04:29But there are other areas, you know, and you said it on your show earlier, you had the Bank
04:34of America CEO on, you know, where actually some of that needs a little bit of patience
04:39to say, we've got to make sure we're doing this right.
04:41And we're not just going to release fully autonomous agents out without any real thought.
04:46But I mean, McKinsey, I mean, you guys are advising a lot of these companies on this
04:49sort of AI build out for lack of a better phrase.
04:52I mean, would you advise them to slow things down?
04:55Would you actually point to anything and say, maybe you shouldn't adopt this particular
04:59agent or this particular practice?
05:01So first and foremost, I would advise them to say, look, what's the outcomes that you're
05:06actually trying to drive here?
05:08What is the benefit that you're going to see?
05:10Maybe it's for a particular workflow.
05:11Maybe it's for an entire domain.
05:13And actually, I think there's a lot more value in focus than there is on sort of proliferation
05:18of AI.
05:19So that's the first thing.
05:20And the second thing I would say within it is responsible AI needs to be baked into, you
05:25know, day one of how you're thinking about it.
05:27It's not the check and balance that happens after the fact.
05:30It's a thing that you are thinking about right the way through every choice that you make.
05:34And that still needs to shift left.
05:35I still have one quick question here, only about a minute left.
05:38And this was another data point in your study about software use.
05:42And it was something like 30-something percent of companies that basically decided to get
05:45rid of at least one type of software because of the AI tools and enhancements that have
05:51come about here.
05:53As you know, investors kind of freaked out about this a little bit, software investors
05:56earlier this year.
05:57Should they be freaking out more?
05:59I'm not sure freaking out is the way forward on any of this stuff.
06:02Should they be cautiously exiting?
06:03So I think what people need to do is take a hard look at their stack and say, look, is
06:08this still fit for purpose for what I'm actually trying to achieve, back to what the outcomes
06:12are that I'm trying to get to?
06:14And that normally results in a whole variety of choices up and down the stack.
06:20But for me, it's not a one-size-fits-all.
06:22Different software for different uses, different ability to link that in with the human
06:27wear, as I said earlier, of your organization.
06:30So not one for freaking out on any particular topic.
06:33All right.
Comments