Post # 88
May 7, 2025
Claire Bodanis
Yesterday, the Falcon Windsor/Insig AI team published ‘Your Precocious Intern’, our research and practical recommendations for how to use generative AI responsibly in corporate reporting, based on our project with FTSE companies and investors. In today’s blog, prompted by the unexpected appearance of generative AI in her personal life, Claire cross-examines her own motives, and invites you to share your thoughts too.
Friday Night Dinner is an important event in the Bodanis household – and no, I don’t mean the popular British sitcom (even though one of the main actors is brother to my son’s headmaster! How very North London…). As regular readers will know, my husband David is Jewish, so each week, we have our own version of Friday Night Dinner, albeit with the drama provided by a cast of my younger Gentile relatives and their entourage.
Last Friday, we had a brand new guest of our own to bring the drama, a young American chap related to a friend of David’s, who’s just moved to London to start his own AI-powered tech business. Very timely, since only that afternoon I’d signed off the vast labour of love that is our research and recommendations for how to use generative AI responsibly in corporate reporting, published yesterday (6 May). Neither David nor I had met him – in fact our invitation to join the gang for dinner was our way of welcoming him to London. Far more interesting, I thought, to meet a bunch of people closer to his own age, than just us fogeys.
Unfortunately, the poor chap had managed to get off on the wrong foot before even appearing on our doorstep, so extending the usual warm welcome required a little effort on my part. Why? Let me share with you our email exchange over said invitation.
I write:
Hello, hello! Welcome to London. I’m usually to be found here – just not much in April. How about Fri 2 May at ours?
Claire
He replies:
Hey Claire, Friday, May 2nd dinner at yours works great!
Anything I can bring?
Cheers.
Hmmm… ‘thank you’ would be nice, but whatever. He seems jolly enough. And then I realise that’s not the end of the reply.
The email continues:
Thanks Claire! Unfortunately I can't make May 2nd. Could we look at [alternative date] instead?
Cheers.
Yes, dear readers, it doesn’t take a genius to work out what’s happened here. I am both amused and, I admit, rather annoyed.
I write:
?? We seem to have two messages, one saying yes and one saying no… you’re not getting AI to draft your replies, are you…?! Haha…
He replies:
Hahaha busted xD I am a yes for Friday May 2nd dinner!
Re the two messages – I started using a tool called Cora which is meant to save you time on your email. One of its features is that for some emails it drafts responses so you can edit. It works pretty well... when I remember to delete the other drafts lol.
I reply:
You need to be careful with that… it’s a good job I’m not important!
We look forward to seeing you in four weeks’ time!
Claire
You will note the use of exclamation marks, carefully added to take the slight sting out of my response. But for someone who’s looking to find backers for a new business, I felt it was a warning worth giving, even if it was, unusually for me, rather on the passive-aggressive side.
Before you get too excited, I’m afraid this blog is not leading up to a mighty Friday Night Dinner showdown worthy of the screen. We had an entertaining and fun evening, along with some good debates about generative AI. Which included hearing our new friend’s view that, in 10 or 20 years’ time, nothing’s going to be written by humans at all, and, despite being very nice and personable, and well aware of the potential issues, the idea didn’t seem to bother him much. But I must confess that I couldn’t get that little niggling feeling of pique out of my head.
Given that the main thing occupying my mind these last few weeks in the run-up to publishing our paper has been where to draw the line about what generative AI should and shouldn’t be doing, I thought it might be helpful to examine this feeling.
Why was I so bothered? Is it reasonable to feel like this? Unreasonable? Am I just old-fashioned and out of touch? If I am, might that mean (eek) the research and recommendations we’re publishing about how generative AI should be used in reporting are also old-fashioned and out of touch?
A possible answer, at least to the first question, presented itself via my junior doctor niece, who said (after our guest had left, when I shared the story to amuse the remainder of the party, who reassuringly also thought it was a bit off): ‘I don’t mind when I get what I think are AI-generated emails from senior consultants – I just assume that they’re really busy and have more important things to do.’
Was that it? We’d invited him to dinner, yet whatever else he was doing was more important than taking the trouble to write his own reply. Was it a case of wounded pride? Partly. But it was more than that: in failing to make that small effort, he’d caused me to waste my time having to find out which of his replies was the one he’d meant to send. So, a perceived lack of respect then?
Turning my mind back to our research discussions, the extent to which generative AI should or should not be allowed to write things on behalf of people was by far the most heated and contentious area of discussion during our research with FTSE companies. Far more, even, than concerns over the accuracy of information; which you might think would be top of reporters’ risk list, given the well-publicised problems of generative AI producing false information and false source references.
Equally, when it came to the investor discussions, their main concern was how to ensure the opinions expressed remained truly those of management and the Board. And even if they were, what did it say about respect for their shareholders if they couldn’t be bothered with reporting and just got a bot to do it?*
But were those sentiments just an age thing? Everyone we talked to was, inevitably, of a similar age, people whose formative educational years were, like mine, spent with pen and paper, let alone word processing and computers. As generative AI becomes increasingly pervasive, perhaps it won’t matter any more. Perhaps we – or if not us, then coming generations who grow up with it – won’t care that others get their bots to communicate on their behalf, rather than spending the time themselves, because everyone’s doing it. Perhaps, as our dinner guest suggested, generative AI writing everything will become the new social norm, and we’ll find other ways to show that we care about each other.
Now hang on a minute. Despite professing not to be bothered about generative AI writing everything, his response when he got found out – ‘Busted xD’ – implied that, on the contrary, it did matter. That in fact he didn’t want me to know that he’d used AI to reply; he wanted me to think he’d written it himself. Why? Presumably because, deep down, he instinctively felt some of the same qualms about how the person on the receiving end would feel about getting an automated response. Particularly when that person – i.e. me – was someone bestowing a favour.
But what of reporting? It’s not interpersonal communication, so surely no one’s going to feel personally slighted by an AI-generated statement? On the contrary… precisely because it is impossible for a Chair or a CEO to have a direct relationship with every individual stakeholder, or even every individual investor, reporting stands as a proxy for that relationship. Reporting gives those who do not have the privilege of personal, face-to-face interactions that essential insight into the minds of management and the Board.
This concept is at the heart of the UK Financial Reporting Council’s requirement that annual reports be ‘fair, balanced and understandable’, perhaps because reporting – the opinion parts at least, if not all the detailed disclosures – is the principal way that Boards and management can ‘show that they care’ about their investors and other stakeholders. And in the absence of any direct, interpersonal form of communication, it’s arguably the only way they can show that they care.
Perhaps I’m not so old-fashioned after all…
…or am I? I’d love to know what you think. Please do get in touch!
PS You may be amused to hear that, just before publishing this blog, we received a very lovely thank-you note, which I am pretty certain was NOT written by AI!
*Incidentally, respondents generally didn’t feel the same concerns about human writers helping the Board and management write their opinions, assuming those writers were properly engaging with the individuals in question. For reasons of length I won’t discuss this here, but invite you to read about it on p20 of the full research paper!