They are still behind OpenAI. They are going to lose this next race as ChatGPT becomes people’s de facto little computer friend they ask every question that comes through their heads. Not to say google will not be a successful company yet, but its search has one foot in the grave.

(http://www.autoadmit.com/thread.php?thread_id=5729736&forum_id=2).#48960303)

I'll admit I could be mistaken. Perhaps its a GIBBON my limited primate nose is sniffing out.

(http://www.autoadmit.com/thread.php?thread_id=5729736&forum_id=2).#48960299)

and at least that garbage code written by mammals eventually works. AI slop won't work without significant human intervention unless you're building something pretty simple.

(http://www.autoadmit.com/thread.php?thread_id=5729736&forum_id=2).#48960341)

This shit makes up basic CLI commands. To think this is better than humans is absurd and falling for the hype/marketing.

(http://www.autoadmit.com/thread.php?thread_id=5729736&forum_id=2).#48960385)

Also I don't care what the test says. I have used the o4 mini high API to do advanced tasks and have seen the smartest humans do the same thing (actually not really because it does things no human could ever do, but I have seen them attempt it). Its not even a comparison. The AI is more intelligent than any human that has ever lived by orders of magnitude that might not even be measurable.

Also have you considered that syntax errors in coding might not be the best test of intelligence?

(http://www.autoadmit.com/thread.php?thread_id=5729736&forum_id=2).#48960409)

And again, your examples are garbage. You are using CLI syntax and "1 million line code" haunted spaghetti fortresses" as your benchmark for intelligence. Congratulations, you've rebranded chronic human dysfunction as a benchmark. What you are really saying at this point is "if the machine doesn't code with trauma and self-doubt it doesn't count". And you do understand there are things that require more advanced reasoning than hobbling together a tech stack right? Or is that really the example we are going here when o4 mini high can process reams of Grothendieck and Lurie in under a second?

(http://www.autoadmit.com/thread.php?thread_id=5729736&forum_id=2).#48960459)

Infinite IQ on display

(http://www.autoadmit.com/thread.php?thread_id=5729736&forum_id=2).#48960465)

https://www.theguardian.com/technology/2025/may/20/almost-half-of-young-people-would-prefer-a-world-without-internet-uk-study-finds

the Uncle Ted effect

(http://www.autoadmit.com/thread.php?thread_id=5729736&forum_id=2).#48960276)

That’s all you’re getting out of this

(http://www.autoadmit.com/thread.php?thread_id=5729736&forum_id=2).#48960364)

Ofc this isn't really that different from people simply googling a topic and posting the first link which is basically how it worked before

(http://www.autoadmit.com/thread.php?thread_id=5729736&forum_id=2).#48960479)

He won't of course but this is the obvious solution. It's going to become a serious problem, people will just spam it on every single post

(http://www.autoadmit.com/thread.php?thread_id=5729736&forum_id=2).#48960485)

In 10 years we'll be dealing with a generation of adults who have never known anything but

(http://www.autoadmit.com/thread.php?thread_id=5729736&forum_id=2).#48960740)

https://www.youtube.com/watch?v=vbWe5k4fFWE

(http://www.autoadmit.com/thread.php?thread_id=5729736&forum_id=2).#48960566)