YT Commenter 2015

3.5K posts

YT Commenter 2015

YT Commenter 2015

@ytcommenter2015

Katılım Nisan 2015
66 Takip Edilen13 Takipçiler
YT Commenter 2015
YT Commenter 2015@ytcommenter2015·
@emollick The whole point of strawberry is to expose that GenAI doesn't really understand. Its pattern matching/interpolation can yield some awesome results on seemingly complicated problems but it's still the case that it can also yield extremely stupid results on trivial problems.
English
0
0
0
214
Ethan Mollick
Ethan Mollick@emollick·
Its funny how much the whole "strawberry" thing, which turned out to be o1-preview, was dismissed as overhyped at launch when it is clear in retrospect that it was way underhyped. A direct line from models unable to do basic math to solving unresolved math problems in 18 months.
English
55
75
1.4K
170.7K
roon
roon@tszzl·
gary marcus, beyond just being plainly & clownishly wrong for years, is a slimy and bad person and you shouldn’t encourage him even as a rhetorical object
English
138
66
2K
152.9K
Yuchen Jin
Yuchen Jin@Yuchenj_UW·
This one won't age well.
Yuchen Jin tweet media
English
98
11
451
95.4K
YT Commenter 2015
YT Commenter 2015@ytcommenter2015·
@MattBruenig @moetkacik This is only true of either OpenAI/Anthropic. A ton of compute capacity doesn't actually seem to be in high utilization. They were so low for Grok/xAI that they are now renting it to Anthropic.
English
1
0
0
59
YT Commenter 2015
YT Commenter 2015@ytcommenter2015·
@MattBruenig @moetkacik This doesn't discuss the impact on residential living at all. All these people who think that data centers are fine should live next to one. The 24/7 noise alone would be enough to drive you crazy.
English
0
0
0
63
Lee Roach
Lee Roach@leevalueroach·
The biggest contrarian signal I watch is “The Kobeissi Letter”. When this engagement bait account starts talking about a stock. I get interested. Biggest contrarian signal in the free market today. You can make a lot of money doing the opposite of whatever this account says.
The Kobeissi Letter@KobeissiLetter

BREAKING: The Trade Desk, $TTD, falls over -20% after posting weaker than expected earnings. The stock is now down -85% since December 2024 in a historic corporate collapse.

English
121
52
1.6K
860.6K
YT Commenter 2015
YT Commenter 2015@ytcommenter2015·
@zck Go live next door to Data Centers. If you are good with the constant noise, utility cost spikes, then maybe people will listen. Otherwise, fuck off.
English
0
0
1
60
YT Commenter 2015
YT Commenter 2015@ytcommenter2015·
@AndyMasley You are comparing twitter shit posts to purported information retrieval system in reliability? The GenAI platform is causing sloppy contents, sloppy/buggy products. Twitter posts don't cause producers to generate sloppy products.
English
0
0
0
132
Andy Masley
Andy Masley@AndyMasley·
idk just going down the list: 1) Yes xAI's Colossus in Memphis and Mississippi are bad. They are extreme outliers in the AI industry for how close they emit to people's homes. 2) AI accelerates climate change in the sense that it uses energy. I think the only way to fairly judge how much something accelerates climate change is to think about it in a per capita way, because otherwise its only sin is that it's popular. Like chocolate bars have more emissions in total than a single yacht, but I think the yacht is more wasteful for climate. Per user, AI adds extremely small levels of CO2. The reason it's having such an effect is that it's being used so much, not that each use is especially bad. 3) Since when do we think of individual bottled water companies as being a unique threat to water? 4) Yes all global AI use emitted more than New York State. Most global industries do. 5) So has the internet. It's bad, but is the answer just not using the technology at all? 6) I've known a lot of people who were basically driven insane by using the internet. It's a real problem. Do we stop using the internet? How often does this actually happen out of the hundreds of millions of people using it every day. 7) Information from this twitter poster is also tremendously unreliable. 8) Only if you use it badly. This is otherwise an argument against books. 9) If I go to a museum and see art I like, and draw something inspired by it that doesn't come close enough to it to actually breach copyright, is that evil? Do I need the artists' permission before doing that? That's what's happening here. Weird for an anarchist to defend IP. 10) It's not using that much water. By 2030, all American data centers may use 1% as much water as irrigated corn. 11) The government also does this with the internet. Should we stop using the internet? 12) If AI is so useless, wouldn't you prefer it if people doing bad things used it? Wouldn't that incapacitate them? 13) The tackiness of art has zero relationship to its politics. Leftists make lots of tacky art too. Fascists can also use photoshop, is this an argument against photoshop? I think these are mostly wrong, but the real threats of AI, like power concentration or surveillance or even extinction are very real. But to fully understand either I think it helps a lot to actually try out the models to see what's up. As it stands, most arguments I see against using AI apply to literally any industry at all.
SUPPORT MUTUAL AID/DIRECT ACTION 🏴 Ⓐ@tigerfoos

I made a zine about why AI sucks I thought people might enjoy looking at it

English
26
45
478
31.6K
YT Commenter 2015
YT Commenter 2015@ytcommenter2015·
@AndyMasley If you have been to neighbors next to datacenters with constant noise running 24/7, you will know why.
English
0
0
1
122
YT Commenter 2015
YT Commenter 2015@ytcommenter2015·
@AndyMasley How much is the tax revenue for local governments? And how much tax cuts do corporations get on equipment? I suspect that the numbers are drastically in favor of the corporations.
English
0
0
0
11
Timothy B. Lee
Timothy B. Lee@binarybits·
It's wild that we are having an environmental panic about data centers — a carbon-free industrial facility that during ordinary operation consumes no natural resources besides electricity and water and emits no waste products besides heat.
English
84
94
892
85.7K
YT Commenter 2015
YT Commenter 2015@ytcommenter2015·
@petergyang Have you used any AI bots that purportedly do troubleshooting? They are utterly useless and sometimes worse than simple document search.
English
0
0
0
599
Natalie Wynn
Natalie Wynn@ContraPoints·
It’s become a standard leftist view that Trumpism was a backlash to Obama not being socialist enough and I… don’t think that’s what happened lol
The Lever@LeverNews

How did America go from Obama to Trump? Maine Senate candidate @grahamformaine tells @davidsirota it wasn’t some great mystery: Democrats bailed out banks, abandoned working people, and let corporate power keep running the party. Go and listen to the full episode over on The Lever's Youtube channel.

English
694
443
6.8K
975.2K
Derek Thompson
Derek Thompson@DKThomp·
Again, if you thought that the odds that AI was a bubble in 2025 were >50% (hi) I don’t know you can look at this and not shift your odds. The best time to flip-flop is when reality changes. A leading AI company’s run rate quadrupling in six months is some kind of change!
Josh You@justjoshinyou13

not a super clean report, but looks like Anthropic has reached ~$45B run rate one month after reaching $30B. this is a continuation of the tripling per quarter/~100x annualized growth rate they've seen this year.

English
45
32
706
112.4K
YT Commenter 2015
YT Commenter 2015@ytcommenter2015·
@binarybits Was it mythos? Or was it that they decided to engage with various teams to look for more security bugs? Maybe they would’ve found more security bugs before without Mythos if they decided to do extensive engagement. You are just taking Anthropic announcements uncritically.
English
0
0
3
429
Timothy B. Lee
Timothy B. Lee@binarybits·
Not a good day for team "Claude Mythos is just marketing hype."
Timothy B. Lee tweet media
English
26
77
852
73.5K
YT Commenter 2015
YT Commenter 2015@ytcommenter2015·
@signulll x.com/theinformation… x.com/HedgieMarkets/… So called compute demands are extremely concentrated in basically a couple of labs.
Hedgie@HedgieMarkets

🦔A Cast AI analysis of roughly 23,000 Kubernetes clusters found average GPU utilization across enterprises sits at 5%, meaning 95% of provisioned GPU capacity is idle. CPU utilization averages 8% and memory 20%. Companies are overprovisioning out of fear of missing allocations rather than sustained demand. The CEO of Cast AI described it plainly: companies are overbuying GPUs out of fear of missing out. My Take Last night Meta, Amazon, Microsoft, and Alphabet all reported earnings beating expectations while simultaneously raising AI infrastructure spend. Meta alone raised its 2026 capex guidance to $125-145 billion. GPU spot prices are up 48% in two months. CoreWeave raised rates 20%. The narrative driving all of that is insatiable AI demand that companies cannot afford to miss. Against that backdrop, 95% of provisioned enterprise GPU capacity sitting idle is a figure I find genuinely difficult to square with the shortage narrative. The price increases and multi-year infrastructure commitments are being driven by demand signals from companies that purchased GPUs on the same fear-of-missing-out logic and are now leaving 95% of that capacity unused. Companies are paying for GPUs they aren't running while bidding up the price of GPUs they claim they urgently need. The hyperscalers booking record backlogs last night are counting commitments from the same enterprises sitting on idle infrastructure, which is a different thing from collected revenue, and the distinction is important when the bill for all of it eventually comes due. Hedgie🤗

English
0
1
1
169
signüll
signüll@signulll·
just ~six months ago so many ppl thought everything was bubble… e.g. too much compute, too much capex, & demand that couldn’t possibly absorb the buildout. but it turns out the ceiling on demand for intelligence is literally nowhere in sight. i mean who could’ve ever seen that a smart entity available on tap is actually insanely useful for ~every facet of human life & even for things we haven’t even thought of that are possible now???
English
134
131
2.3K
298.7K
YT Commenter 2015
YT Commenter 2015@ytcommenter2015·
@binarybits @GaryMarcus Investments are also a form of gambling. In this case, they require probably the biggest returns ever in terms of raw monetary value.
English
0
0
0
18
Timothy B. Lee
Timothy B. Lee@binarybits·
@GaryMarcus Maybe they are making big investments now (and made big investments in the past couple of years) because they expect it to be useful in the future? Investments are forward-looking almost by definition.
English
2
0
9
1K
Gary Marcus, MIT PhD and NYU Professor Emeritus
👇 This is an incredible admission by Jensen Huang. Effectively he is saying that AI, for all the hype (some from himself) wasn’t really useful until late 2025. Let that sink in. This means practically everything everyone said before was more or less bullshit. And it raises a question: *how useful* is it now, and how much it that utility worth? $10B a year? $50B? It can’t just be a little useful to justify the trillions in infrastructure. It has to maybe be the most lucrative technology that ever existed to justify those investments. So far it ain’t even close to that.
tae kim@firstadopter

Jensen: "this all happened in the last several months. AI in the last several months became useful. That’s the big idea" "GPU consumption is going through the roof. Even GPUs we sold four or five years ago now are rising in price faster than good wine." $NVDA

English
39
48
294
31.7K
YT Commenter 2015
YT Commenter 2015@ytcommenter2015·
@jackclarkSF Rather non stop PRs, you should focus on fixing your broken business practices and all the sloppy bugs in your websites and CC.
English
0
0
0
34
Jack Clark
Jack Clark@jackclarkSF·
I've spent the past few weeks reading 100s of public data sources about AI development. I now believe that recursive self-improvement has a 60% chance of happening by the end of 2028. In other words, AI systems might soon be capable of building themselves.
English
289
499
3.5K
1.6M
YT Commenter 2015
YT Commenter 2015@ytcommenter2015·
@mcuban You just realized that GenAI hallucinations are unsolvable?
English
0
0
0
4
Mark Cuban
Mark Cuban@mcuban·
I’m coming to the conclusion that the biggest challenge for Enterprise AI, and AI in general , as of now, is that it’s still impossible to make sure that everyone gets the same answer to the same question, every time. Which is a great response to the doomers. AI doesn’t know the consequences of its output. Judgement and the ability to challenge AI output is becoming increasingly necessary, and valuable. Which makes domain knowledge more valuable by the second. Am I wrong ?
English
1.9K
445
6.2K
1.5M