One curious and unexpected outcome of this AI experiment is that we must concede that it probably does a better job than us (as opposed to being just an expedient and quick but mediocre solution). We have experienced in real time what the population is probably mulling over in their sub-conscious – “what then, is the point of us (apart from actually managing portfolios)?”  Happily, we have come through the other side, and we’re ok with it (the ‘population’ should be relieved). At this point it is worth making a distinction between extractive AI (creating content based on known inputs) and generative AI (creative insight generated from the internet ‘collective’ along with the occasional hallucinatory experience). We are obviously using the former, which is a much more robust approach, which involves asking the AI to find some material (which we vet), adding some more material that we have come across during the week, and then asking it to write about it. Just how well it writes is a bitter pill to swallow for anyone that takes any pride in how they write, but this affront is softened by the fact that it purports to write in our style – the best of both worlds (and at least our feelings are spared!) One of the more tortuous, but sometimes rewarding, elements of this type of market writing is finding the common threads and relating it to portfolios or our current area of focus. It turns out the AI (in this case a quite well-designed front end to ChatGPT4) is very good at this too and it is a little spooky how it seems to find relevance beyond what we were aware of prompting it for. One traumatising aspect of the process has been the challenge to add something more thoughtful to the AI’s commentary – the end result being a burst of productivity on Monday morning followed by several days of thoughtful procrastination.

On the other hand, AI is by design much less capable of finding flaws in the consensus which it has so ably parsed, and that is probably where we should be spending more of our time. So, in summary (population take note), we think AI, as a market content generator, raises the bar and also forces us to make sure we are actually adding value and not painting by numbers.      

With all that in mind, we will be sending out an AI generated summary more promptly on Monday mornings, followed by a more thoughtful piece during the week, which will usually include a video.

As it happened, last week was uneventful and markets have been more or less flat for the last 10 days, with the exception of the UK, which rallied on the news that inflation was not as high as expected (though still higher than most places), plus some of the economic data has not been quite as dire as has been expected. You can find a pretty good AI generated summary of what happened last week here – the gist of it is that recent trends have continued but without too much drama, with economic data looking weak, although supply constrained housing markets around the world have held up, the consumer (particularly in the US) has been resilient, while inflation pressures appear to be receding, and wage inflation remains relatively constrained. Yesterday the US Federal Reserve raised rates one more time by 0.25%, very much as expected, and Jerome Powell managed to say nothing that would upset markets about their future intentions. On the face of it, this could look like the immaculate disinflation and soft landing that markets have been hoping for. Meanwhile in Australia these trends may look slightly worse, with retail faltering, especially given today’s unexpectedly weak retail sales number that has put the market on the back foot. The Australian inflation print earlier this week surprised on the downside, which was ostensibly a good thing that the market liked quite a lot, but perhaps also not unrelated to today’s weak sales. This may be a salutary reminder to be careful what we wish for and highlights the fact that this soft landing is an incredibly narrow path to stay on, and a lot can go awry betwixt and between.

‍