Transcription
The last big tech bubble burst at the start of this century, and we may be heading that way again. The kind of investment wave in AI we've seen is like probably nothing ever before in history. Hundreds of billions of dollars are being spent on automating workplaces. We have this amazing technology. However, we're not seeing adoption fully yet in every pocket of the economy. Only 1% of CEOs have a fully formed AI strategy. With such high stakes, will businesses see a return on investment?
I'm Isabelle Barrett. I lead the FTS working at Grand, speaking, presenting, and writing about management, leadership, and workplaces. In this series, I'll explore some of the most pressing issues around the future of work and talk to senior leaders about how they are making work better.
"3-5 years from now, I think things will look quite different for everyone."
I'm here at the Charter Workplace Summit in New York, in rooms filled with senior leaders from some of America's biggest companies. These are the people tasked with AI rollout and preparing the workforce for the skills needed for the future.
"Last is David."
"Every 6 months a new model is dropping. Every 6 months something shifts within the marketplace where you have to stay up to date. With AI, we're still in like the very, very, very early days of everything happening. We have this amazing technology with the promise of productivity enhancing gains. Roughly 10% of companies are fully starting to integrate AI into their processes. But there's going to be years of this happening. We have to figure out exactly how we can use it and where it makes sense to use it."
A staggering amount of investment has been made in AI over the last few years, and it now accounts for a 40% share of US GDP growth this year. Over 75% of businesses worldwide are using generative AI in at least one function. But despite this, a study by MIT Media Lab found that 95% of Gen AI pilots in the workplace failed.
I spoke with editor-in-chief of Charter, Kevin Delaney, about the state of AI rollouts in industry.
"Think about how AI is different from humans."
"Companies are adopting AI at two separate speeds. You have the tech companies who are actually quite far along, to the point where they think of AI agents as co-workers. On the other hand, you have companies that are still getting their heads around what AI adoption means. And these are the companies that are still trying to get their employees to use ChatGPT or Claude. A lot of them are not seeing gains in productivity at this point. So you have these two extremes."
"So we hear a lot about, um, the need to upskill the workforce for AI. What does that actually mean? Are people actually doing it, or are they just letting people get on with it?"
"People are trying to figure out what exactly that means. And I think part of the challenge is that we don't actually know what the ideal worker skills will be in 3 years or 5 years as AI is rolled out more pervasively. There's a lot of discussion about: Is the ideal worker in a more AI-deployed environment someone who is a real specialist in a field, or is it someone who is a generalist who kind of knows a little bit about the business and how business operates, and who can communicate clearly, and knows enough to be able to check what the AI is bringing back?"
"So we need a lot more experimentation and possibly failure."
"Yeah. And so that's uncomfortable for leaders, too. To be comfortable with failure is something that you are not generally taught in business school. Failure generally is something that executives are allergic to encouraging in their workers."
After a day of off-the-record discussions, panels, and big-picture sessions, what's emerged is that there's no clear path forward for GenAI at work. It's still all to be decided. The reimagination of work.
Leaders have spent billions on preparing for an augmented future, but for what gain? So, at the FT, we wanted to look at how is this rollout actually going, and what are companies saying about how they're using AI. And so, we did this massive analysis looking at, um, S&P 500 companies in the US. Um, we went through thousands of earnings reports and, um, regulatory filings, and the s- the results were quite surprising. Um, in earnings reports, CEOs would often say, you know, 'AI is amazing. It would bring incredible productivity gains, a Cambrian explosion of innovation,' things like that. Um, but then in the filings, which, to be fair, tend to be more measured and, um, risk-averse, no one really had anything concrete to say of how they're actually using it. And, uh, in those filings, the risks outweighed the benefits very, very clearly.
If you look at the S&P 500 index, it's obviously going up, but a lot of that growth is driven by seven big tech companies. And the other companies on the S&P 500 haven't necessarily grown that much when they've said they use AI. AI use is often phrased in their filings as being something quite abstract. Um, they talk about productivity but don't really offer any concrete examples of how they're using it. Coca-Cola is one example where in their earnings reports they raved about how they're using generative AI to transform their business. Um, but in their filings, the only example they could give was using generative AI to create a Christmas ad. It's definitely a mixed bag.
The growth of AI has led to a boom for consultancies and learning platforms who are keen to show business how to harness the powers of AI at work. I visited the HQ of AI upskilling platform Multiverse and met with their CEO and founder, Euan Blair.
"So what are the ways in which companies, I guess your clients, are engaging with AI skills? Are they hesitant? Are they all in? How is it, what does it look like?"
"So, I, I, I think it's, it's almost the kind of polar opposite of hesitant. The kind of investment wave in AI we've seen is like probably nothing ever before in history. So the big challenge a lot of organizations are facing is how to turn kind of potential AI gains into actual realized AI gains. And that's where the kind of training gap comes in because what a lot of people are doing with AI at the moment is the equivalent of having an iPhone and just using it to send text messages and make calls, right? They're missing out on loads of the capabilities that these tools actually have. So we've seen a lot of companies spend a lot of money on AI and, really a lot of money, and there haven't been, right, particular productivity gains that I'm aware of."
"What's, what's the, where's this gap? What's the gap?"
"We've seen accounts teams, for example, um, process invoices 50% more quickly and with half the number of errors because of introducing AI. We've seen, um, software engineering teams increase their speed of shipping code, uh, by 75% in some cases. Those are big, tangible things that do actually have an impact. One of the reasons we're not seeing gains at the kind of big macro level yet in terms of economic growth is this sort of training and capability gap. Because with previous versions of software, it was often deemed enough to go and invest in the technology and then over a period of several years, people would figure out how to use it and where to use it, and everything would be okay. The difference this time is the inherent capability of the systems is so much greater. You need a lot of training to be able to kind of fundamentally change the way you work, but also the amounts being spent are so much greater. So the stakes are higher, and that kind of creates this, this perfect set of conditions where people realize the people who spend the most on AI are not the ones who are going to win. It's going to be the people who have the most AI-enabled workforce, and that's the kind of space Multiverse is playing in."
Everyone feels like they're behind the curve when it comes to AI, and they all feel like they're not doing enough and could be doing more. And that is kind of creating this sort of, um, it's not even a hype cycle, but it's a just a desire to kind of do more, faster. So when you think about the financial gain of AI, a lot of that money is flowing into tech companies, AI companies, management consultants, and companies adopting AI aren't necessarily seeing those magical financial gains that they were promised. But it's worth bearing in mind that it's still really early on. Um, it's really early in the deployment stage of these technologies. Just a few years ago, they were still in the lab. And so we have to be patient. But obviously, the question is, how long do we have to wait? Obviously, businesses are hoping that these use cases and gains will come sooner rather than later.
The number of people turning to commercial AI platforms on a daily basis has been astronomical. The rate of adoption for ChatGPT alone outpaces the rise in use of the internet when it was first launched. But the gulf between work-related and personal usage is growing. So what you often see are these shadow use cases where official, uh, corporate AI initiatives are often untouched or unused, and people just use the AI tools they like. And this is often because there, there hasn't been necessary communication between leadership and staff about what they need and what kind of tools they actually want. But different rules apply at workplaces, right? Workplaces often have sensitive information, or accuracy really matters, and so you have to pay attention to the fact that these models often do make factual mistakes, and that could be really embarrassing or even catastrophic for an organization. So, every organization needs to be thinking about this and thinking about how these tools apply to them and what they want their employees to know about how to use them.
Some of the biggest challenges that businesses face are that they just aren't ready for this digital transformation. To use AI well, you need good structured data, good cyber defenses, and perhaps most importantly, AI-literate staff. I went to Google's newest campus in New York to meet Amanda Brophy, director of Grow with Google. It's Google's professional training arm and offers courses to businesses and individuals on how to use AI.
"What's your advice for leaders who have maybe a cohort of staff who are still very skeptical of AI or slow to adopt?"
"I think you need to find how to make the AI work for that specific person in their role and what they're doing. What makes AI so powerful is when you can translate it into what you are doing today, and now that's specific to you. So if a marketer is trying to use AI, and we're helping them figure out how to use this to write social captions for their social media posts; for customer service, to think about how they use this to write responses back in a way that's polite when someone's getting upset and it's escalating. Making it custom to that person and role is when you actually see the real benefits. And so being able to test that for you is what allows that skepticism to go away and see the real benefit from it."
"One of the big problems with AI rollout is that people aren't really getting trained. So what do you say to employers?"
"You need both the technology and the training, right? You need the tools and the training. It's an 'and,' not an 'or.' And so what we're finding is just rolling out the technology isn't enough. We have a course, the Google AI Essentials course. And what we've seen is that being able to teach people how to use the technology, how to prompt, and make sure that they're using it in an effective and reliable way helps them to get to use it every day, to upskill and reskill. What I think makes AI different is it's not learning about it. It's, you have to use it and do it. You have to have the daily practice to make it a regular habit in the work that you do. It's one of those ones that you need to have the intrinsic interest to be able to see the value of AI in the day-to-day of your professional and personal benefits, and the employer needs to be able to deliver and have this available for employees so that people are consuming this information for their, for the company."
"And what's your best tip for anyone watching this who wants to get better with AI in their job?"
"You need to be able to prompt the AI effectively to make sure you get the desired output that you want. Highlighting pieces like who's the audience you're trying to reach? What's the goals and the context? What's the reference materials? And so being able to prompt AI effectively is critical to get the output that you will then see to make this a regular habit and the efficiencies that you want."
"So do you think journalists make good prompters?"
"I bet we do."
"I think you make excellent prompters because you're good at the questions. It's exactly what it is. You understand who the audience is, what the questions are. I think journalists are excellent prompters."
Perhaps not surprisingly, the tech sector has been an enthusiastic AI adopter. I met with Cisco's UK and Ireland CEO, Sarah Walker, to see how it's working for them.
"So internally at Cisco, it's a tech company ahead of the curve. What does AI usage look like generally internally here?"
"Really, really broad spectrum. So if I think of it in terms of our product development, um, things like our Webex platform have AI agents built in, and they do some fabulous things which have made my life a lot easier and more efficient. We've also then got some really great platforms that we use as, as employees. There's different levels of adoption of that. As you can imagine, some are super proficient, some still are trying to get to grips with with what that means. But that's where adoption becomes key because for, for us to really capitalize on the efficiencies that those investments, um, can and should deliver, um, our next task is, you know, how do people adopt that and, and make that a part of their kind of DNA and how they operate on a daily basis."
"From talking to people, there's a kind of, you know, people bring in AI systems and then they don't really monitor adoption. How can leaders get over that?"
"Well, first of all, you have to lead by example because my team will never adopt those sorts of platforms if I'm not talking about it and using it myself. So, we did a masterclass actually with our senior leadership team across the UK. And I speak really, really positively about pro-workforce and pro-AI. It's not an either-or, and using AI doesn't mean that at some point in the future your role will be replaced by it. This is about using these applications to say, how do you become more efficient in the things that you can and should automate. And candidly, it's human nature to want to find a quicker and more efficient way to do things. We've always been like that. Just because it's now called AI or that's more kind of broadly known, we shouldn't be, we shouldn't be fearful, um, of that. But it is a, a common mistake that businesses make, that thinking just because you've got the applications or the opportunity that adoption will follow. Everyone should definitely try these tools. They're a lot of fun to play around with, and that's the quickest way to learn how these might work for you or how they might not work for you. You have to use them for use cases where the tools are actually beneficial instead of expecting it to be some sort of magic wand that can fix all problems. And so currently, we're operating in the fact that this all will work and it'll lead to amazing things in the future. But if that were to change, if this were a massive bubble that were to burst, um, the reality is that a lot of these AI experiments, only the use cases that actually work and that bring benefits to employees will stay. Everything else, I can't really see surviving."
The challenge of AI rollout in workplaces doesn't have a one-size-fits-all solution. Businesses need input from staff, but equally, those staff need support and training from their leaders if any of us are to realize the financial and productivity gains that AI promises. I'm old enough to remember when the internet rolled out in the mid-1990s, and it seems to me we're at a very similar early stage of the cycle with GenAI. There's a lot of boom and bust to come, and with it, disruption, and I hope, excitement at work.