This is pretty silly. Your memory obviously gets a little worse as you age, with most of the noticeable decline coming in very late age. But the artificial measures/definitions of things like "fluid" intelligence are mostly useless. Just pulling up one of the studies cited in the article which is supposed to measure the "reasoning" aspect of "fluid" intelligence, presents a huge host issues immediately [1].
Aside from the lack of randomization, you have obvious validity problems. The interpretation of nebulous words like "reasoning" as being accurately measured by e.g. accuracy on Raven matrices (construct validity?) and younger participants having been primed by recent test-taking experience while real-world reasoning skills aren't really reflected - it's all quite specious.
Real-world decisions are value-laden and constraint-laden! "Intelligence" does not mean "maximizing abstract pattern detection". If you keep your brain active with a wide range of creative, interesting problems, you will be fine apart from neurodegenerative diseases, which have real effects.
That's exactly how intelligence in measurement setting is defined. That's how the word is defined. What's interesting is that it correlates with so many unrelated real world outcomes. And other definitions and measurements do not.
You're apparently alluding to "IQ" without saying it. But once you get above "low IQ" performance on these tests, their predictive value trends to zero. The correlation is very one sided (as explained by people like Nassim Taleb). Also, when I say "real world problem solving", that doesn't mean "income". A predictor can look powerful even if it's not measuring real-world skills, as long as institutions use it to distribute opportunities and labels that later show up as "job success".
I never heard of the institution that distributes opportunities based on the result of IQ test. I only heard about IQ test for police and for military.
I'm not sure if that's what you mean by "one sided", but while you can find high IQ people in all walks of life it's very easy to find them in high achievers. And it goes way beyond lack of low IQ people. Even average IQ people are rare there.
Right. If the "slop" is truly "90% as good" and that 10% actually matters to people, then they won't use the slop. If the 10% doesn't matter, there's probably a reason for that.
If that's the case, then business results should get worse, and management should notice this. If business results don't get worse, then either 1) it's actually more than 60% as good, or 2) it doesn't matter to the business's bottom line that the result is only 60% as good instead of 80% as good, and management made the right decision.
>business results should get worse, and management should notice
This is a common oversimplification that results in an enormous amount of waste and bad products/services. Lots of causes and effects are too disconnected to see or too hard to measure. In addition to looking at metrics, good business leadership most also act like a human (which is a depressing thing to have to say): Use common sense; like good things; dislike bad things.
This reads as if written by someone who has never used these tools before. No one ever tries to "fit" the entire project into a single context window. Successfully using coding LLMs involves context management (some of which is now done by the models themselves) so that you can isolate the issues you're currently working on, and get enough context to work effectively. Working on enormous codebases over the past two months, I have never had to remind the model what it changed in another file, because 1) it has access to git and can easily see what has changed, and 2) I work with the model to break down projects into pieces that can be worked on sequentially. And keep in mind, this the worst this technology will ever be - it will only get larger context windows and better memory from here.
What are the SOTA methods for context management assuming the agent runs with its tool calls without any break? Do you flush GPU tokens/adjust KV caches when you need to compress context by summarizing/logging some part?
And they shouldn’t raise red flags. 99.9% of the symptoms any individual experiences in their life are not life threatening. If you got scans and interventions at every symptom, you’d end up in debt and with iatrogenic harms. Until we can get more accurate data from our bodies more easily, this is the reality. The good news: there’s so much room for innovation and progress in medicine.
Not angry nor scared, I value my hard skills a lot, I'm just wondering why people believe religiously everything AI related. Maybe I'm a bit sick with the excessive hype
There's no fear (a bit of anger I must admit). I suspect nearly all of the reaction against this comes from a similar place to where mine does:
All of the real world code I have had to review created by AI is buggy slop (often with subtle, but weird bugs that don't show up for a while). But on HN I'm told "this is because your co-workers don't know how to AI right!!!!" Then when someone who supposedly must be an expert in getting things done with AI posts, it's always big claims with hand-wavy explanations/evidence.
Then the comments section is littered with no effort comments like this.
Yet oddly whenever anyone asks "show me the thing you built?" Either it looks like every other half-working vibe coded CRUD app... or it doesn't exist/can't be shown.
If you tell me you have discovered a miracle tool, just some me the results. Not taking increasingly ridiculous claims at face value is not "fear". What I don't understand is where comments like yours come from? What makes you need this to be more than it is?
It's because they don't have a substantive response to it, so they resort to ad hominems.
I've worked extensively in the AI space, and believe that it is extremely useful, but these weird claims (even from people I respect a lot) that "something big and mysterious is happening, I just can't show you yet!" set of my alarms.
When sensible questions are met with ad hominems by supporters it further sets of alarm bells.
I see way more hype that is boosted by the moderators. The scared ones are the nepo babies who founded a vaporware AI company that will be bought by daddy or friends through a VC.
They have to maintain the hype until a somewhat credible exit appears and therefore lash out with boomer memes, FOMO, and the usual insane talking points like "there are builders and coders".
Claude and Codex are CLI tools you use to give the LLM context about the project on your local machine or dev environment. The fact that you're using the name "ChatGPT" instead of Codex leads me to believe you're talking about using the web-based ChatGPT interface to work on a large codebase, which is completely beside the point of the entire discussion. That's not the tool anyone is talking about here.
Aside from the lack of randomization, you have obvious validity problems. The interpretation of nebulous words like "reasoning" as being accurately measured by e.g. accuracy on Raven matrices (construct validity?) and younger participants having been primed by recent test-taking experience while real-world reasoning skills aren't really reflected - it's all quite specious.
Real-world decisions are value-laden and constraint-laden! "Intelligence" does not mean "maximizing abstract pattern detection". If you keep your brain active with a wide range of creative, interesting problems, you will be fine apart from neurodegenerative diseases, which have real effects.
https://pubmed.ncbi.nlm.nih.gov/30211596/
reply