Another highlight from the actual report is a massive increase in attempts to build AI in-house rather than buy, which highlights existing systems' inability to generate value. We can't find any use case for Clippy 2.0 as part of our existing software but, but the investors (and my bosses) might get spooked if we don't sound like we're on the cutting edge of this tech that everyone says is revolutionary. In the context of 70-90% of software projects failing in whole or in part I can only expect this to go well.
this post was submitted on 03 Dec 2024
33 points (97.1% liked)
TechTakes
1481 readers
393 users here now
Big brain tech dude got yet another clueless take over at HackerNews etc? Here's the place to vent. Orange site, VC foolishness, all welcome.
This is not debate club. Unless it’s amusing debate.
For actually-good tech, you want our NotAwfulTech community
founded 1 year ago
MODERATORS
I've also been involved with projects that were retro-activly renamed to AI. Implemented a new oldskool search algorithm? That's AI now! Simple decision tree with static weights? AI!
Nobody wants investors or customers to think you're doing a bad job not putting AI in stuff, even if it doesn't really help or isn't a use case for anything related to AI.
Some of that is driven by privacy concerns though, at least in my case. In-house solutions means the clients potentially sensitive information is staying in-house as well.
The first internal combustion engines were invented in 1794, while the first production motor vehicle was in 1885. Even if technology exponentially increases, there could still be some time before we find the best application for it.
THAT BEING SAID, the application of the internal combustion engine makes me think about a Douglas Adam's quote: “In the beginning the Universe was created. This had made many people very angry and has been widely regarded as a bad move.”
before ICE there were steam engines, used for locomotives as soon as it became practical, and before that these were used as stationary power sources. these things were immediately useful
cars were made really, really practical only with invention of tetraethyllead, which made everyone exposed stupid and aggressive, and also made petrol manufacture cheaper spurring car-centric transportation which has many harmful consequences even today
genai is much closer to tetraethyllead in few aspects - pollutes internet for years to come and spews disinformation, increases carbon dioxide emissions without good reason and paves way for SV nepo babies into even more influence
Early ICEs also saw immediate value in static applications like water pumps or printing presses. It's not like there was a century of people tinkering with the thing trying to make it work without any purpose. It rapidly showed value and then was iterated upon and improved until it became practical to use in new ways e.g. being small enough and powerful enough to power a car.
I figure there's a small chance that the AI bubble ends up acting more like the UK railroad bubble in that after the dust clears and the overproduced infrastructure gets shut down we're left with the best rail lines in Europe, but I'm not going to bet on that unless and until the salesmen stop trying to sell everyone a rail line between their bedroom and the bathroom.
In the famous locomotive competition where Rocket beat Novelty (or was it the other way around?), other locomotives also participated. Some broke down and one was disqualified for containing a horse instead of a steam engine. Feels like there are lots of hidden horses today, and they are rewarded instead of disqualified.
Hidden horses is too good of a phrase to leave buried here
We lost 'Mechanical Turk' as a descriptor for AI because it's literally the name of the service they use for labeling training data. 'Actually Indians' is still on the table.
pollutes internet for years to come and spews disinformation,
I wonder if future search engines will have an option to ignore results from 2022 to 20?? to avoid woeful AI content?
...which made everyone exposed stupid and aggressive...
Funnily enough, this seems to be true for gen-ai, too!
both are fundamentally heavily optimized for corporate profit too. TEL was cheaper than alternatives available then, like adding alcohol to gasoline or using more BTX fraction. The problem is that the former cuts into oil company profits and the latter requires lots of heat to manufacture. Today we have better processes and blending stocks like alkylate that avoid some toxicity of BTX and all of TEL, but these would be never developed out of corporate goodwill, they were forced to by law. Idk what purpose spicy autocomplete has other than hyping up stocks
N-rays were discovered in 1903, clearly it's just early days