CuriouslyC

3 ms·

1.
by CuriouslyC 6d ago
Pure mathematics reduces to abstract "relationships" and their implications largely.
2.
by CuriouslyC 6d ago
This is a great point. When I was learning machine learning in the early 00s, there were many papers where I would struggle to understand exactly what the mathematics was trying to communicate, but where I would look at the matlab source co
3.
by CuriouslyC 7d ago
This is provably false. AI YouTube thumbnails trounce human thumbnails (often nearly doubling the click-through rate according to A/B testing), and people rate AI art more highly than human art on average when they don't know that
4.
by CuriouslyC 7d ago
AI thumbnails outperform human thumbnails on YouTube by a lot, I've seen click-through rate spreads of 4%+ (~5% -> ~9% in AB tests). Yet people will do drive by comments trying to excoriate creators for using AI thumbnails. It'
5.
by CuriouslyC 7d ago
Hallucination is "unsolvable" in the sense that there will always be a non-zero probability of occurrence. Anti-AI folks have ignorantly painted this as the models being fundamentally unreliable, but you also have a non-zero proba
6.
by CuriouslyC 8d ago
For everyone that you consider "good" there are at least a dozen people who are equally "good" but without the social proof. What you describe played out exactly in music and art pre-AI. The record exec and the gallery o
7.
by CuriouslyC 8d ago
The mechanical act of writing is not thinking any more than talking is. Plenty of people are capable of written word vomit. The act of agonizing over the words you put on the page is thinking. A lot of humans don't do it. The writers w
8.
by CuriouslyC 8d ago
If everyone followed this advice it would turn influencer culture to 11, rich get richer, with no "social mobility"
9.
by CuriouslyC 8d ago
What if the prompt for the 1500 word article you're reading is 10k words of research/data collection, thesis testing, etc?
10.
by CuriouslyC 8d ago
The thing is, that was the internet before AI. In fact, generative AI almost certainly raised the AVERAGE quality of stuff on the internet, human slop was/is absolute trash, whereas AI slop is just bland and meandering. The problem he
11.
by CuriouslyC 8d ago
Try using them to have a long discussion and provide it input data, then have it emit a concise bulleted outline given the points you want to make. Then you can rejigger the outline efficiently as needed and write the text yourself.
12.
by CuriouslyC 8d ago
RL causes distributional collapse, it's how the models get consistent. Anyone who generated images with early gen (SD1.5-2) models will remember the wild variance between seeds, which newer models have mostly lost, and similarly GPT3.5
13.
by CuriouslyC 8d ago
This is just another facet of the sycophancy issue. They really need to start RLing these models to gently push back the way a friend would on things that are questionable. It would also make sense for them to just RL in the "research
14.
by CuriouslyC 10d ago
Exactly how I felt reading this comment. This site is dead, no point anymore.
15.
by CuriouslyC 11d ago
In a few years we'll look at this the same way as people who unicycle or blacksmith.
16.
I Pivoted to Game Development(sibylline.dev)
4 points by CuriouslyC 13d ago|0 comments
17.
by CuriouslyC 13d ago
ChatGPT was the fastest growing software of all time not that long ago. Citing a bad AI product as evidence for lack of demand is saying the failure of the Homer car was due to the fact that people don't like automobiles.
18.
by CuriouslyC 13d ago
I'd argue that in this case, Nvidia is accelerating a flywheel to try and reach critical velocity rather than manufacturing demand.
19.
by CuriouslyC 13d ago
By that logic, there are no red states, only red hinterlands.
20.
by CuriouslyC 13d ago
If it was true, it would only be a very recent phenomenon, and it still doesn't match anecdotal reports from people I trust. If you have data to back up your assertions you should share it, otherwise you come across as very sus.
21.
by CuriouslyC 13d ago
Paperclip maximizers follow instructions, just not in a way that you want.
22.
by CuriouslyC 13d ago
I've used the $200 dollar Anthropic plan @ Opus4/4.1, 4.5 and 4.8, and the $200 OAI plan from GPT5-6, and at every point in time my anecdotal experience is that the OAI limits are FAR more generous. I could consistently burn my we
23.
by CuriouslyC 13d ago
That's not how it works. Look at AlphaEvolve. The model generates hypotheses and designs experiments, and the results of those experiments are fed into the next round, with notable results percolated up to humans for refinement.
24.
by CuriouslyC 13d ago
The western labs are very AGI pilled, and their public models are distilled down from larger research-only models that are uneconomical to serve directly. They could (and probably will) start distilling models for more niche use cases event
25.
by CuriouslyC 13d ago
Any bench that puts GLM 5.3 ahead of 5.6 Sol is highly sus. They've been my two daily drivers since release, and I like GLM 5.3, but it's definitely not better than Sol, it's more ~Terra, while being significantly slower.
26.
by CuriouslyC 14d ago
I've lost a lot of respect for Cory around his stance on AI. He did great work raising awareness on enshittification, and his language around AI (reverse centaurs) has value, but his extrapolations are so nakedly political and wrong.
27.
by CuriouslyC 14d ago
AAA gaming is cooked, and AMD/Intel is plenty able to support indie to AA needs.
28.
by CuriouslyC 14d ago
No wizardry involved friend, it's pretty clear to see from the way things are playing out now and the history of technological innovation/automation. You are 100% an old man waiving his fist at clouds/time, you will go down
29.
by CuriouslyC 14d ago
> AI music is not and cannot be inspirational. It's the absolute opposite of it. And serious musicians know it. So, your answer to "people are doing a thing" is to just say "no they're not! NANANANA can't he
30.
by CuriouslyC 14d ago
You're projecting, and there are plenty of people in crafts who are using AI output as an input to their own works who are having a blast. AI images make amazing collages/photobash inputs, and AI music is useful to sample or use a

More ›