The thinking feeling thinking machine
This is a fascinating bit of thinking, about what might start to happen as we use LLMs to make product/buying decisions.
"In that way, it creates a funny marketing paradox where we think that consumers are purely emotional beings who fail to think rationally, even though Herbert Simon proved that their emotional approach was actually economically rational. On the other hand, these models, which we think of as perfect embodiments of logical thinking, are actually far more emotional, aiming to give us what they think we want, rather than actually acting rationally."
The important bit. re-explained so I'm sure I understand.
The general assumptions at large are, probably, currently:
- Humans make emotional decisions
- Computers make rational decisions
Therefore communications are going to have to change in favour of rationally persuasive information. LLMs after all can consume and process a lot of it.
BUT
LLMs/AIs aren't trying to make good decisions they're trying to please us.* So, if we want to influence them we have to give them what we think they think we want.
And what they seem to think we want is stuff that looks rational and logical. They've ingested so many seemingly persuasive bullet points and charts and information-in-threes that they think that's how human decision-making works. They're not concerned about rational benefits and are more interested in things that look like rational benefits.
(Remind me to tell you sometime about ICD)
It's a bit like the frequent phenomenon of someone saying Hey This AI has Solved PowerPoint - Look At This Fantastic Presentation. And then showing you some awful, awful slides. AI doesn't do good PowerPoint it does what the internet thinks is good PowerPoint. (And the internet is bad at knowing that)
Similarly, presumably, AI won't make good decisions it will make what the internet thinks are good decisions. And it won't provide useful information, it will provide what the internet thinks is useful information.
All of which makes the whole AI/SEO dance a particularly stupid one right now. There's not much point trying to optimise content to serve something so opaque and weird and Hawthorne-Effect-y.
*And, of course, they're probably not processing information and making decisions, they're just word-predicting.
