What lies ahead, in 2025?

I had the good fortune to catch-up with a dear friend and professional associate yesterday; someone I have not had the chance to interact with for nearly nine months.

Tony and I chatted about all manner of things: his new barn (read: man cave and office); our exploits with Rosa (read: 1951 MG Y-type); geopolitics; ChatGPT; and more besides. What was fascinating was that we both found ourselves chatting as if the last time we spoke was yesterday. ​Before we knew it, some 75 minutes had passed by. ​My father told me that this is a good thing; a sign of true friendship.

One aspect of our conversation that piqued my attention was Tony’s investigations around artificial intelligence and board reports—or, more specifically, his application of large language model tools to discern and make sense of board reports. The rapid progress over the past twelve months is a sight to behold. Tony summarised his experiments and findings. Did you know that if you feed ChatGPT a set of board papers and ask it to summarise the key points, including nuances and appropriate questions to ask in a board meeting, the likelihood of the responses being both insightful and relevant is high? You can also use it to discern whether directors have read and understood the board papers! I have been a sceptic about the application of AI tools for some time but, on the strength of what was outlined, I’m ready to believe ChatGPT (or Claude, or other) can be a real boon for directors struggling to make sense of large data sets. While context eludes ChatGPT (and all other LLMs), and meaning and reasoning too, the direction and pace of travel seems to be reasonable. Certainly, progress is rapid.

I went to bed after our call pondering a plethora of options, including whether board directors might be supplanted by machinery in future. Of this, I am doubtful. But where LLMs could be quite valuable is to distinguish between lights in the distance: those that are sunlight at the of the tunnel, and those that are a train heading towards me at great speed.

And so, with 2025 underway, is your board ready for what lies ahead? Can it, for example, confidently distinguish between [sun]light at the end of the tunnel and a train headlight? Has it carefully considered options having read widely, invoked various tools including AI tools and debated options; or, does it remain reliant on what management feeds up in the board report? To rely on management reports as the sole source of ‘truth’ is not smart; it never has been.

PS: this is Rosa: