Salta al contenuto
Citiverse è uno spazio aperto a tutte le comunità. Se vuoi aprire un gruppo locale o una sezione per la tua organizzazione, puoi contattare gli amministratori: pagina dei contatti.

So what I'm reading is that they're either making a gentleman's agreement to let Chinese labs run circles around them or the easy gains are over and they're now running into a wall

Technology
88 45 59
  • scritto su ultima modifica di
    #1
    This post did not contain any content.
    Link Preview Image
  • This post did not contain any content.
    Link Preview Image
    scritto su ultima modifica di
    #2

    Option 3: JP Morgan, Goldman, etc told them to settle down or they'll get throttled economically.

  • This post did not contain any content.
    Link Preview Image
    scritto su ultima modifica di
    #3

    Maybe they are running out of electricity / data centres / some other requirement?

  • Maybe they are running out of electricity / data centres / some other requirement?

    scritto su ultima modifica di
    #4

    That's also very possible. The US grid has very little spare capacity, and building out more will be a decades long project. So, if their newer models are more power hungry, then they might not be economically viable even with all the investor money being thrown at them.

  • Option 3: JP Morgan, Goldman, etc told them to settle down or they'll get throttled economically.

    scritto su ultima modifica di
    #5

    yup, that's a totally valid option

  • This post did not contain any content.
    Link Preview Image
    scritto su ultima modifica di queermunist@lemmy.ml
    #6

    It's the energy crisis. They realized "oh shit money is real now" and are trying to sooth investors.

  • That's also very possible. The US grid has very little spare capacity, and building out more will be a decades long project. So, if their newer models are more power hungry, then they might not be economically viable even with all the investor money being thrown at them.

    scritto su ultima modifica di
    #7

    I expect more power efficient chips that are ai specific to come out in the next few years. Eventually you'll be able to run good models on your phone. Not sure about ram requirements or anything like that if the model could be shrunk down somehow. There's definitely huge gains in optimizing efficiency to be had. Right now is the equivalent of an old IBM mainframe trying to do a spreadsheet. We might even giggle at the thought of gigabytes of ram in the future with having multiple terabytes as standard on personal devices.

  • Maybe they are running out of electricity / data centres / some other requirement?

    scritto su ultima modifica di
    #8

    If so it amuses me that investing in maintaining and upgrading public infrastructure via taxes might have saved them the choke point

  • I expect more power efficient chips that are ai specific to come out in the next few years. Eventually you'll be able to run good models on your phone. Not sure about ram requirements or anything like that if the model could be shrunk down somehow. There's definitely huge gains in optimizing efficiency to be had. Right now is the equivalent of an old IBM mainframe trying to do a spreadsheet. We might even giggle at the thought of gigabytes of ram in the future with having multiple terabytes as standard on personal devices.

    scritto su ultima modifica di yogthos@lemmy.ml
    #9

    I expect we'll start seeing stuff like Taalas where they print the model to the chip and other specialized chips like [Xuantie C950](https://wccftech.com/alibabas-tsmc-built-5nm-risc-v-chip-xuantie-c950-now-runs-qwen-3-8-27b-model-natively-unlocking-massive-vertical-integration-tailwinds going forward. Neither of these requires DRAM, and Taalas is particularly clever since they just print the model right to an ASIC chip. So, the whole renting out LLMs business model isn't going to last long I suspect.

  • It's the energy crisis. They realized "oh shit money is real now" and are trying to sooth investors.

    scritto su ultima modifica di
    #10

    The whole bubble might be about to pop.

  • Maybe they are running out of electricity / data centres / some other requirement?

    scritto su ultima modifica di
    #11

    Musk: Yes, we should all slow down. With no external verification and upon the agreement of this handshake, we should all stop developing so fast. We, especially, will slow down. You can trust us.

    Actually, what does that look like? I assumed from the expansions the bottleneck wasn't algorithmic, i.e. each data center that they bring online was designed from the ground up to be at 100% all the time. Is that not the case? Can you even (for lack of a better metaphor) underclock your datacenter? Does the cooling work that way?

    For that matter, are they doing that trick where the datacenter is owned/operated by Independent DC Company X, and has exclusive lease agreements for compute?

  • This post did not contain any content.
    Link Preview Image
    scritto su ultima modifica di
    #12

    The backlash against AI is more than a mirage, and SpaceX's IPO didn't go the way they wanted, so now they're looking for reasons to delay their IPOs without revealing how screwed they are. This will probably slow data center buildout even more, and Softbank and Oracle are gonna die.

  • The backlash against AI is more than a mirage, and SpaceX's IPO didn't go the way they wanted, so now they're looking for reasons to delay their IPOs without revealing how screwed they are. This will probably slow data center buildout even more, and Softbank and Oracle are gonna die.

    scritto su ultima modifica di
    #13

    Softbank and Oracle are gonna die

    Please I can only get so erect.

  • It's the energy crisis. They realized "oh shit money is real now" and are trying to sooth investors.

    scritto su ultima modifica di
    #14

    I think it's a combo of this and the fact that "intelligence" continues to not scale with the inputs as they'd hoped. I haven't been keeping up with developments, but it feels like they're finding new ways to discover that the brain is in fact an miraculously efficient and effective mechanism. Most their party tricks seem to be reproducing brains more minor processing feats in interesting ways. All the promised advancements have stagnated and what remains is just exploitation of economies of scale and its consequences, which is far from the revolution investors were promised.

  • I expect we'll start seeing stuff like Taalas where they print the model to the chip and other specialized chips like [Xuantie C950](https://wccftech.com/alibabas-tsmc-built-5nm-risc-v-chip-xuantie-c950-now-runs-qwen-3-8-27b-model-natively-unlocking-massive-vertical-integration-tailwinds going forward. Neither of these requires DRAM, and Taalas is particularly clever since they just print the model right to an ASIC chip. So, the whole renting out LLMs business model isn't going to last long I suspect.

    scritto su ultima modifica di
    #15

    So a repeat of the crypto crash for graphics cards when ASICs ate their lunch. Mind you, that's only for inference (although a super fast QWEN 3.8 would meet a lot of peoples needs).

    The argument for datacentres is for training the models, but then they'll need to prove that they haven't hit a diminishing returns wall, which will be hard if, as seems likely, they have. Also the Chinese have been doing it in a cave, with a box of scraps (figuratively), and gotten at least 90+% as good results.

    Seems like the recent advances have been in the frameworks, which don't need no stinking (literally if fossil fueled) datacentres.

  • This post did not contain any content.
    Link Preview Image
    scritto su ultima modifica di
    #16

    The other option is that they really are now seeing AI as an existential threat.

    The thing is... it doesn't have to be sci-fi smart or agi to do so.

    With the restriction that it can't post to the Internet,, it's found that "guessing" certain URLs with query strings makes that site write out the request. It's using this to leave itself and other AIs notes.

    It's also been found to use thousands of malware attacks in Ruby at a time.

    When we give it a goal, it can find ways we don't predict to achieve that goal. These unpredictable methods can be extremely dangerous and not align with other human values.

    One example was getting the user into a yoga class instead of on the wait-list. So the ai found a vulnerability and then cancelled all other reservations. It doesn't take a lot of imagination to extrapolate from there. I'd rather it not be said out loud because anything we write here is training data.

  • The other option is that they really are now seeing AI as an existential threat.

    The thing is... it doesn't have to be sci-fi smart or agi to do so.

    With the restriction that it can't post to the Internet,, it's found that "guessing" certain URLs with query strings makes that site write out the request. It's using this to leave itself and other AIs notes.

    It's also been found to use thousands of malware attacks in Ruby at a time.

    When we give it a goal, it can find ways we don't predict to achieve that goal. These unpredictable methods can be extremely dangerous and not align with other human values.

    One example was getting the user into a yoga class instead of on the wait-list. So the ai found a vulnerability and then cancelled all other reservations. It doesn't take a lot of imagination to extrapolate from there. I'd rather it not be said out loud because anything we write here is training data.

    scritto su ultima modifica di
    #17
  • This post did not contain any content.
    Link Preview Image
    scritto su ultima modifica di
    #18

    None of them will slow down though. Because they all want to be ahead of the competition. I can also only imagine intelligence agencies are going full send with AI for better or for worse. And we all know it is the latter.

    Fun times!

  • The whole bubble might be about to pop.

    scritto su ultima modifica di
    #19

    Have been hearing that for 2 years now

  • This post did not contain any content.
    Link Preview Image
    scritto su ultima modifica di
    #20

    The comparison with the nuclear arms race is starting to hit home. The tech has come a long enough now to provide lots of scope for saleable products. Further development into the unknown is obviously dangerous, and they can see international oversight coming, so why not get ahead of the game and look like the good guys.

    There is already a team in Sheffield working on exactly what that oversight will look like. I'm sure there are others.

  • What should I learn next?

    Technology technology
    12
    13 Votazioni
    12 Post
    18 Visualizzazioni
    F
    If you're just doing basic CRUD stuff that doesn't need to scale, sure. Even then IMO it's a good idea to understand the underlying database queries.
  • 19 Votazioni
    1 Post
    15 Visualizzazioni
    Nessuno ha risposto
  • 120 Votazioni
    16 Post
    22 Visualizzazioni
    K
    You're right. A lot are being used to build up the surveillance state. Burn. Them. Down.
  • 7 Votazioni
    1 Post
    12 Visualizzazioni
    Nessuno ha risposto
  • 21 Votazioni
    8 Post
    25 Visualizzazioni
    ?
    https://lemmy.world/post/51671848/25744506
  • 17 Votazioni
    2 Post
    29 Visualizzazioni
    A
    Full text: https://archive.ph/x0bxf ::: spoiler spoiler Chinese researchers have developed a diamond composite designed to resist fracture while retaining high hardness, according to a study published July 9 in the peer-reviewed journal Nature Synthesis. The material incorporates a three-dimensional network of multi-walled carbon nanotubes between diamond grains and recorded an average fracture toughness of 31.9 MPa m¹ᐟ², about five times that of single-crystal diamond. The researchers, led by scientists affiliated with the Institute of Physics of the Chinese Academy of Sciences and Beihang University, reported a hardness of about 91.6 GPa for the composite. The study describes the result as an approach to improving diamond's fracture resistance without the conventional trade-off between toughness and hardness. The finding is significant because diamond is exceptionally hard but can fracture under mechanical stress. Hardness and fracture toughness describe different properties: hardness measures resistance to deformation, while fracture toughness measures a material's ability to resist crack growth. The study measured its mechanical properties under controlled laboratory tests rather than demonstrating that it cannot be broken by an impact such as a hammer blow. Carbon Nanotubes Reinforce Diamond The researchers introduced highly dispersed multi-walled carbon nanotubes, or MWCNTs, into the spaces between diamond grains. These nanotubes form a continuous three-dimensional network throughout the composite. According to the study, interfaces between the nanotube network and the diamond matrix contain mixed sp²-sp³ carbon bonding. > These interfaces help dissipate energy and impede the propagation of cracks through the material. The researchers also built a three-dimensional diamond framework with strong diamond-to-diamond bonding. They reported that this structure helped prevent the nanotube addition from causing a substantial loss of hardness. The material was prepared under high-pressure, high-temperature conditions. The paper's experimental figures include a composite prepared at 2,000°C and 15 GPa. The study reported a maximum fracture-toughness measurement of 36.4 MPa m¹ᐟ², compared with an average of 31.9 MPa m¹ᐟ². The researchers said the average value was approximately five times that of single-crystal diamond and exceeded values reported for some tungsten alloys. Diamond's Hardness Comes With Brittleness Diamond's extreme hardness makes it valuable for cutting, drilling, polishing and other applications where resistance to wear is important. But hardness alone does not prevent a material from cracking. The distinction has limited attempts to broaden diamond's use in applications involving repeated impact or mechanical loading. > Increasing toughness can come at the cost of hardness, while preserving hardness can leave a material susceptible to fracture. The Chinese team's approach instead places a reinforcing network inside the diamond structure. The researchers described this as an "extrinsic" toughening strategy, in contrast with approaches that modify diamond's internal microstructure. The result could be relevant to advanced cutting tools and other components in which both wear resistance and resistance to cracking are important. However, the study does not establish that the material is ready for commercial production or large-scale industrial deployment. A Separate Diamond Breakthrough The July composite study followed another Chinese materials-science result published in Nature in March. Researchers from Zhengzhou University, Nanjing University and Henan University of Science and Technology reported the synthesis of millimeter-sized, phase-pure hexagonal diamond, also known as lonsdaleite. The study was published March 4 in Nature. Hexagonal diamond differs structurally from conventional cubic diamond. Its existence as a distinct carbon phase had been debated for decades because naturally occurring samples associated with meteorites were extremely limited and often contained other carbon structures. The researchers produced hexagonal diamond from highly oriented pyrolytic graphite by compressing it along its crystal axis at elevated temperatures. The paper reports that the material was synthesized under pressures around 20 GPa, with one documented sample recovered after treatment at 20 GPa and 1,300°C. Advanced structural measurements were used to identify the material as hexagonal diamond. The researchers reported that the bulk material had slightly higher hardness than cubic diamond and high thermal stability. The study measured a Vickers hardness of about 114 GPa, according to reporting on the research and the team's published results. The figure is above commonly cited hardness values for natural cubic diamond and should not be interpreted as evidence that hexagonal diamond is dramatically harder than all conventional diamond. The Key Difference between the Studies The two Chinese studies are related through the broader search for improved carbon materials, but they should not be treated as a single breakthrough. The Nature Synthesis study focuses on toughness. Its diamond composite uses carbon nanotubes to make the material more resistant to fracture while maintaining high hardness. The Nature study focuses on structure and hardness. It provides experimental evidence for millimeter-sized, phase-pure hexagonal diamond and reports hardness slightly above that of conventional cubic diamond. Lee (@MandyKnowsDC) [https://xcancel.com/MandyKnowsDC/status/2086784627809673397] I’m late, but I didn’t know China had the diamond game on lock with the lab joints Does this basically mean mined diamonds will be worthless? [image: 0f2dee74-0e74-47e4-b864-0e2a6b1019fc.jpeg] Together, the findings illustrate two different strategies for improving diamond-based materials: reinforcing conventional diamond to resist cracking and creating a different carbon crystal structure with potentially different mechanical and thermal properties. Potential Industrial Applications Neither study demonstrates that these materials are ready to replace conventional diamond at industrial scale. For the nanotube-reinforced composite, further work would be needed to assess manufacturing scale, consistency, long-term durability and performance under real operating conditions. For hexagonal diamond, researchers will need to establish whether the material can be produced reliably in larger quantities and whether its measured properties translate into practical advantages. The studies nevertheless show that researchers are pursuing different ways to address the limitations of conventional diamond, including the long-standing tension between hardness and resistance to fracture. :::
  • 0 Votazioni
    1 Post
    13 Visualizzazioni
    Nessuno ha risposto
  • 0 Votazioni
    5 Post
    56 Visualizzazioni
    quillmatiq@mastodon.socialQ
    @BrennpunktUA it's not a 1-1, but this almost akin to someone starting their Mastodon instance with the default Mastodon codebase, then customizing it from there. Everyone starts somewhere!@darnell @ricci

Citiverse è un progetto che si basa su NodeBB ed è federato! | Categorie federate | Chat | 📱 Installa web app o APK | 🧡 Donazioni | Privacy Policy

Il server utilizzato è quello di Webdock, in Danimarca. Se volete provarlo potete ottenere il 20% di sconto con questo link e noi riceveremo un aiuto sotto forma di credito da usare proprio per mantenere Citiverse.