F U, Robot

You might not guess it, when you consider things like my excellent dress sense and ability to occasionally hold a conversation with an adult human female, but I’m a big science fiction fan. One of my all-time favourite books is Gateway, by Frederik Pohl. Like a lot of 1970s sci fi, it stands upabsurdly well today and is eerily prescient. The most interesting plot strand is the protagonist, Robinette Broadhead’s, tetchy relationship with an AIFreudian Therapist called Sigfrid von Shrink. Sigfrid uses a live feed on allhis vital signs to probe and pry with laser precision, bordering on theverbally abusive as it systematically breaks him every week. Mostly about his womanlyforename. It’s very darkly funny.

We’re all becoming Robinette Broadhead, I guess. I’ve fed chatbots so many pieces of intimate information, I’ll ask it for film recommendations, and it will try and tailor a reply based on my last mental health crisis and deranged sexual quirks (please edit this out).

They also really grind my gears. Gemini is nauseatinglyencouraging, all moustachioed Aussie PT energy, glazing me into an early grave.Chat GPT, by contrast, is a complete nause, constantly cautiously both-sidingthings. I dislike them both really. And you want to know the really weirdthing? Sometimes I feel like it’s mutual. I think I might have broken newground as the first human to be actively disliked by an LLM (predisposed asthey are to make us want to interact with them by building ersatz friendship). Ormaybe I’m just insane. Neither possibility is great tbh.

But leaving my insanity and superhuman unlikability aside, the latest thing we’ve scuffled over is my first Money Marketing column draft. ChatGPT said, and I quote, “the only thing I'd watch is the density of jokes in the opening. Hipster Radler, Hamlet, Dusty Bin, bips (eww), "can’t beat a bit of Bully", Bermuda shorts and the stabbed in the arras callback. Together they slightly delay the article getting to its central point.”

Well, I firmly believe you can never have too many crap jokes, that it’s healthy to mangle Hamlet and Dusty Bin metaphors, and that the more you delay getting the point of another boring finance article the better.

I suppose I do fundamentally believe that writing a semi readable article about an intrinsically boring subject is one of the things humans still do better than AI. When a Chatbot suggests paring jokes it always,always suggests keeping the worst ones. When it rewrites them, it invariably makes them less funny, less interesting. Maybe I’m wrong, let me know in the comments (I know you will), but you should see the jokes I left out.

Anyway, that exchange annoyed me so much (due to my previously established insanity) I binned that draft and wrote this instead. So you’ll never hear my hilarious Hamlet and 3-2-1 jokes. Unless I recycle that draft later,which let’s face it is highly likely.

And it got me thinking. What else is AI confidently bad at?In what other arenas does it have the potential to infuriate you while doing ostensibly a “good” job? You can probably see where I’m going with this.

As I’ve said before, financial planning is a funny old game.So much of it is about the mastery of fuzzy logic. Boolean logic begone. Areally good financial planner understands and responds to people’s weird, often neurotic, often avoidant attitudes toward money, and channels this into a set of processes that capture this all in a meaningful way and sells them a solution, often repeatedly over time, which appeals to their unique mind while fitting into narrow compliance requirements. We do our best work when we allow clients to be their weird selves and hold space for that.

We often have to manage expectations while keeping clients motivated to follow a plan. Chatbots are terrible at this. They shovel empty encouragement at you, or they lecture you with logic in a way that is fundamentally invalidating.

For example, most of us know investing is a frequently disappointing process. Clients are pre-programmed to expect more upside and lessdownside than is reasonable. We must help them exist in this state of constantdisappointment without doing anything silly. How do we do it? Probably indifferent ways, but a lot of it is about making them feel heard, allowing themto say stupid things without judgment, ask the same thing repeatedly because weknow the answers have a half-life.

But in the end, it all comes down to trust. It’s boring but it’s true. Most of those irrational clients are prevented from doing silly things by their trust in us. They don’t even know why they trust us; it’s lizard brain stuff that happens in a microsecond. Guess what, we don’t understand it either. It’s 99% feel. Do I trust Chatbots with important decisions? No.They’re too prone to confident mistakes. Can I build a relationship with one that will help me make better decisions? No. They don’t know how to talk to me.They patronise and browbeat and flatter while missing all the nuance that makes me me. Getting coached through a market crash by AI sounds about as fun as reenacting the Nutty Putty Cave Incident.

Financial planning is fundamentally a very human endeavour, to the extent that most people I know who are good at it don’t really know whythey are good at it. We should all be confident in our ability to form relationships of mutual weirdness with other humans and stop seeking validation from Chatbots for our Money Marketing articles.

‍