Toggle sidebar Toggle sidebar
Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.

Anthropic Launches Claude Opus 4.8 With Gains in Coding and Honesty

We just don’t need this. We need educated, intelligent, compassionate, caring and considerate humans who will look after each other and the planet. Will the tech bros and politicians ever realise this? No. Because all they want is $ and power. AI (arguably things like SM too) is just another way, especially for the tech bros, to get that. Stop using this stuff. Stop ceding $ and power to those who have enough of both a billion times over, and care little if anything for the rest of humanity and the planet.
 
  • Like
  • Disagree
Reactions: purplefox and iMac The Knife
neuropsychguy said:
There is a whole area of philosophy devoted to this discussion. Some people start with the premise that intelligence requires a biological substrate. Others acknowledge intelligence might arise from silicon or other materials going into computers -- however, some define intelligence in a manner that does not allow current models to be defined as intelligent. Others argue that it currently is intelligent: https://www.nature.com/articles/d41586-026-00285-6

It's an interesting area of discussion.
Click to expand...

Interesting topic. IMO, we might have to take a step back and really question whether we can accurately define intelligence or we can identify high intelligence if we encounter it.
 
  • Like
  • Haha
Reactions: Agent007, neuropsychguy and johnsawyercjs
MacMan988 said:
Interesting topic. IMO, we might have to take a step back and really question whether we can accurately define intelligence or we can identify high intelligence if we encounter it.
Click to expand...
People seem to fall into the trap of thinking “intelligence” is somehow “human” - the concept of intelligence gets anthropomorphised, so we have this very stupid situation where AIs are judged on mimicking personality.

I’d a colony of ants intelligent? Is a plant intelligent? Both have agency, both can adapt without external instruction to changes in environment, both can solve challenges without external instruction. Both can react in complex ways to stimuli that have not been encountered before.
 
Last edited:
  • Like
Reactions: MasterControlProgram and 01cowherd
I am truly impressed with what you get with Claude for the price you pay.
Every day, productivity gains are truly astonishing. It's deja vu when we first got "the internet"

Also, using a tool, paying for it, and not seeing ads is worth the price of admission.
 
  • Like
Reactions: MasterControlProgram and Schtibbie
Schtibbie said:
How old are you? You an edgy teen? Grow up man. Generative AI is not always a slop machine. You don’t like it, don’t use it.
Click to expand...
Quantity over quality. That’s pretty much the definition of slop.

Generated images all have the same feel, code has ridiculous bugs, writing comes out bland. I’ve tried. After fighting with Claude to get the code right, I gave up and found it faster to just write the code myself. The only people who don’t consider it slop are those who don’t know any better.
 
  • Like
Reactions: ElectricPotato
Happy_John said:
No, because it is a tool, not a person. The responsibility lies solely with the person using it.

If you accidentally killed someone with your car or rifle, the car or rifle would not be responsible, you would be.
Click to expand...
Isn't there some responsibility on the company that produced and sold the tool if it doesn't work as intended and as it was advertised?
 
  • Like
Reactions: 01cowherd
judgeFire said:
I find Claude exceptionally useful, and the more humane LLMs appear, the easier it is to treat them like some random person with an opinion on everything, rather than an actual superintelligence. Like asking my uncle about anything in the 80s. He sure made a convincing case, but…
Click to expand...

ok: I now feel old 🥸
 
Schtibbie said:
Oh we have an expert here! How intensively do you use or develop AI that we can take your word that there’s no progress?
Click to expand...
I have a degree in Information Systems, that helps.

Benchmark scores systematically overstate capability, and the mechanisms are well documented. Items leak into pretraining, confirmable via n-gram overlap if you would like to check it yourself, and reconstructions like GSM1K drop frontier models by about 13 points, with drops correlating to verbatim-reproduction propensity: a memorization signature, not difficulty. Post-cutoff accuracy also falls below matched pre-cutoff items. Plus Goodhart: labs optimize toward public metrics. Score = capability + contamination + optimization bias.

TL;DR: Labs tune models toward public benchmarks, through data selection, fine-tuning, and design choices, so scores rise without proportionate gains in the underlying capability the benchmark was meant to proxy.

You are welcome. For further reading, let me refer to corresponding work by Hugh Zhang, Manley Roberts, Oscar Sainz, Yonatan Oren et al.
 
Last edited:
  • Like
Reactions: iMac The Knife, ItWasNotMe and BSDnostalgia
You must log in or register to reply here.
Register on MacRumors! This sidebar will go away, and you'll see fewer ads.
Top Bottom
Back