Working It’s guide to AI at work

Talk of how generative artificial intelligence is shaking up our work life is everywhere — from its ability to automate mundane everyday tasks, to producing personalised avatars to take our place in meetings.

But how much does the average worker yet understand about the opportunities and risks for their careers and how can they get the most out of the rapidly evolving technology?

The question prompted the Financial Times’ Working It podcast to produce a practical three part mini-series about AI at work: what it can do, what it can’t do — and what might happen in the future.

The podcast’s host and producer have distilled what they learned into five top insights. Listen in full to the series here, or wherever you get your audio.

1. AI helps with succession planning — and career progression

Everyone knows generative AI is trained on the data it is fed and, when it comes to its use in recruitment, that has downsides. Much attention has been paid to potential hiring biases and other grim (let’s hope, unintended,) consequences. If the data contains human biases, these will be reflected in potentially discriminatory AI decisions.

But Chano Fernández, co-chief executive of Eightfold, an AI-based talent platform, points out the ways the technology can enhance recruitment practices.

“Usually large [tech] companies tend to think about succession management for the top 200 executives,” he says. AI expands that — we might say democratises it — suggesting succession plans for lower-ranked employees and helping companies identify candidates for vacant roles from a wider pool.

The software can highlight vacancies to a large number of people — but a twist is that it can also show candidates which key skills they are missing for a particular vacancy. That gap might prevent workers from getting the job this time round, but, Fernández says, the AI helps workers to look at the future in a more structured way. It will show them how to acquire the skills to be eligible for that level of job but it will also, he suggests, “expose different career paths, different journeys that [they] could have as a professional”.

This is useful for businesses preparing to hire — and also prompts more imaginative career planning for workers.

2. Prompt engineer will be an in-demand job

Prompt engineering — the words, queries and requests that are fed into an AI tool to generate the desired output — is going to be the in-demand job of the future (so those humanities degrees may not be in vain). Generative AI is only as good as the instructions humans give it, so we need to develop our communication skills — and it will learn with us.

Marcus du Sautoy, the Simonyi professor for the public understanding of science at Oxford university, tells us how that is shaping up: “I'm not particularly good yet at prompting [AI] to go in the directions I would like to push it. [But} I can see that there are already people who are practically artists of the prompt. They are almost writing poetry into the AI, which elicits . . . incredible visual or oral responses.”

Dan Sherratt, vice-president of creative and innovation at the design agency Poppins, gives a practical example of the basics of prompt engineering: “It still is a computer. It still responds to commands. And that is still, essentially, when you break that down to its simplest form, ones and zeros. You have to give it commands in order to receive the right answers. For example, you could ask it specifically to create a photograph that was shot on a specific kind of film, with a specific camera at a specific time of day. So recreating images is far more impressive than just using it like a stock photography search engine.”

Prompt engineering is a skill — not just a set of instructions — and companies are increasingly likely to need to hire people who are good at it.

3. AI is already taking our work — but not the good stuff

When we asked our podcast guests about AI models taking away their work, they said broadly the same thing: AI is (or will soon be) taking some of it. But it is not taking away the parts that necessarily matter to them. This reminds us of the well-established practice of “job crafting”, where workers are allowed to ditch the bits of their job that they don’t like in order to get more meaning and purpose from the parts they value. In the past, this probably meant an unfortunate colleague was landed with your boring data inputting or admin tasks. Now generative AI can do it.

To ease the demands on her time, Iliana Oris Valiente, an executive at Accenture Canada, turns to her “digital twin,” Laila. Trained on years of Accenture case studies and research reports, Laila “can and absolutely will” attend meetings in Valiente’s stead. That means tasks get done more quickly, as Valiente is freed up — and she can dodge repetitive conversations about previous projects. “By the time you’ve had [that conversation] 15 times, it’s really not that interesting,” she says.

The podcast team’s bigger question is around the advisability of allowing digital assistants, and AI generally, to take on more of our work. We asked Madhumita Murgia, the FT’s AI editor and author of Code Dependent: Living in the Shadow of AI, to give us her perspective on what the tech can — and can’t — do for us at work. There are two aspects to this: first, the human one. We may not feel comfortable with an avatar coming to our meetings. And what does it say to the team if a manager just never turns up in person? A digital twin may even look like the boss — Laila looks like Iliana, for example — but perhaps that makes it even weirder?

And, more importantly, AI just makes stuff up. As Murgia says: “I think it’s going to take a lot more behavioural change for us to have these twins or assistants do our jobs, because they’re not fully accurate. They’re riddled with errors. This is inherent, actually, to how generative AI systems behave. They make things up.

“So for us to say, ‘is it OK for me to have a digital twin take notes in a crucial meeting that I’ve got to act on’? — that’s a big jump from a human perspective. I’m not convinced that we’re all just going to divest responsibility on that.”

4. Human-made creative work will be a luxury, high-status product

We are often happy to pay more for a coffee made in a trendy shop, where the barista puts their signature pattern into the foam, rather than choosing a (slightly) cheaper drink ejected at the touch of a button. That analogy can help explain the potential future of creative work. We will pay extra for the cachet and superiority of something crafted by hand (or by mind, in this case).

For jobs, that means individual, high-value creative work involved in industries such as design and marketing is not likely to be replaced by AI anytime soon.

Dan Sherratt says his company uses AI “almost exclusively in the pitch and proposal process” and would not use it in the actual execution of creative work. “It’s really, really good for conceptual work and it’s not quite as good as you’d expect it to be yet at producing finished files.”

He notes that “higher value is [already] applied to things that were made with harder work [in industries such as fashion] . . . You buy a custom pair of shoes from an Italian shoemaker who’s got a family business hundreds of years old, versus a mass produced sweatshop sneaker that has no intrinsic value, or storytelling, to be had with it. So, creatively, that could be where we’re heading.”

5. AI can help introverts to shine

The use of AI in recruitment is already fairly widespread, with tools available to, for example, screen job applications and help candidates write CVs. Ali Ansari, founder of recruitment service micro1, goes one step further, offering an avatar interviewer in its AI-based system for software engineers. He says this can be particularly beneficial to introverts and people with neurodiverse conditions such as autism, for example, who can find face-to-face interviews (whether online or in person) intimidating.

Does an avatar ease the tension? Yes, says Ansari. “We’ve seen a lot of comments about that, candidates basically saying. ‘I’m an introvert and this was really great. I was able to talk about the sort of architecture of some back end system or the design of some front end system, without being nervous that I’m talking to another human’.”

He says this sort of digital interview can improve the experience for candidates and “eventually be much better than that of a human technical interview”.

As new generations enter the workforce, having spent their childhoods gaming, avatar interviewers are likely to feel like a natural progression.