Matisyahu's Blog

I was watching the following video and thought about it in context to the post I made before entitled ‘I’m sorry, I don’t believe they’ve changed’ (link) that I made a few days ago where I pointed out that I don’t believe that the videos of people regretting their vote will actually translate into a long term change of those people’s politics.

The original post I made wasn’t just in reference to Trump supporters but also independents (those who don’t identify with either party) who voted for Trump but are having regrets. I talked about the idea that far too many Trump supporters appear to vote based on vibes but I also wonder to what extent are we not picking up everyone – how many Trump supporters are stuck in the sunken cost fallacy vs staying for the sense of community, belonging, identity being defined through ‘the struggle’, the comradery that they get being part of a movement; the tendency for movements to continue past their used by date.

Regarding the whole AI hype frain, put me in the category of being an ‘AI Hater’ given the amount of time and resources wasted on what has become yet another attempt by Silicon Valley to look for the next growth market – metaverse failed, 3d televisions and phones failed, the ‘touch first’ vision that Microsoft put out with Windows 8 failed, the belief that the AR/VR headset would become the next smartphone etc. Here we are and now Silicon Valley is convincing the coke snorting micro dosing types on Wall Street that the next ‘big thing’ is just around the corner and all they need is a few hundred billion more and then something magical will happen.

One of the major promises of the ‘AI revolution’ is that it will allow us to overcome the limitations of our current forces (which is causing persistently high inflation) to herald an age of abundance where money would cease to having meaning aka a post scarcity society. Sounds great in theory but we’ve seen promises like this is the past regarding other technological leaps – sure, there were improvements in productivity but the degree in which the technology improved productivity never really lived up to the hype. The current crop of LLMs are already hitting their limit with each increase in the size the improvement in accuracy decreases meaning at some point we’re going to hit some sort of plateau and thus will need to develop something better than LLMs.

You’ve heard about the big lay offs in the press but what isn’t covered in the press are the businesses that months later reverse their decision and try to rehire all those people they had laid off. What also isn’t being reported on is how well it performs when compared to humans – how accurate are they when compared to a human, when it comes to code aka vibe coding is the output unreadable verbose slop resulting in technical debt or will it actually improve product quality and productivity. All the test so far that have been done has resulted in underwhelming results and one could argue that the results would be even more underwhelming once these services are no longer subsidised and are charged at the cost of delivering it. Unlike the dotcom bubble where there was actually something useful being sold, I think at the end of the AI bubble there will be very little in the way of scraps that can be used to build the next push forward for the technology sector.