bicboss
LightNews LightNews
bicboss.bsky.social
bicboss
@bicboss.bsky.social
13 followers 42 following 120 posts
Posts Replies Media Videos
Reposted by bicboss
edzitron.com
Clammy Sammy is out of ideas! The Era of Smiles begins soon!
drewharwell.com Drew Harwell @drewharwell.com · Oct 14
Sam Altman went from “AI will cure cancer" to "ChatGPT porn" in less than a month
ChatGPT boss predicts when AI could cure cancer We made ChatGPT pretty restrictive to make sure we were being careful with mental health issues. We realize this made it less useful/enjoyable to many users who had no mental health problems, but given the seriousness of the issue we wanted to get this right. Now that we have been able to mitigate the serious mental health issues and have new tools, we are going to be able to safely relax the restrictions in most cases. In a few weeks, we plan to put out a new version of ChatGPT that allows people to have a personality that behaves more like what people liked about 4o (we hope it will be better!). If you want your ChatGPT to respond in a very human-like way, or use a ton of emoji, or act like a friend, ChatGPT should do it (but only if you want it, not because we are usage-maxxing). In December, as we roll out age-gating more fully and as part of our “treat adult users like adults” principle, we will allow even more, like erotica for verified adults. 12:02 PM · Oct 14, 2025 · 1.2M Views
October 14, 2025 at 6:19 PM
Reposted by bicboss
matthewhughes.bsky.social
My latest newsletter is about how generative AI doesn't just need money to survive, but faith.

Every time we see another big genAI foul-up, that faith will waver, and its end becomes all the more likely.

whatwelost.substack.com/p/generative...
Generative AI’s Impending Death By A Thousand Rake-Smacks
Give it enough rope...
whatwelost.substack.com
October 8, 2025 at 4:50 PM
Reposted by bicboss
edzitron.com
Every single AI data center is a toxic investment, with most of its value tied up in GPUs that will be multiple generations behind by the time these things turn on - and then die in 3-5 years.

All to build for AI demand that doesn't exist.
edzitronswheresyouredatghostio.outpost.pub/public/promo...
Actually, wait - how long do GPUs last, exactly? Four years for training? Three years? The A100 GPU started shipping in May 2020, and the H100 (and the Hopper GPU generation) entered full production in September 2022, meaning that we’re hurtling at speed toward the time in which we’re going to start seeing a remarkable amount of chips start wearing down, which should be a concern for companies like Microsoft, who bought 150,000 Hopper GPUs in 2023 and 485,000 of them in 2024. Alright, let me just be blunt: the entire economy of debt around GPUs is insane. Assuming these things don’t die within five years (their warranties generally end in three), their value absolutely will, as NVIDIA has committed to releasing a new AI chip every single year, likely with significant increases to power and power efficiency. At the end of the five year period, the Special Purpose Vehicle will be the proud owner of five-year-old chips that nobody is going to want to rent at the price that Elon Musk has been paying for the last five years. Don’t believe me? Take a look at the rental prices for H100 GPUs that went from $8-an-hour in 2023 to $2-an-hour in 2024, or the Silicon Data Indexes (aggregated realtime indexes of hourly prices) that show H100 rentals at around $2.14-an-hour and A100 rentals at a dollar-an-hour, with Vast.AI offering them at as little as $0.67 an hour. This is, by the way, a problem that faces literally every data center being built in the world, and I feel insane talking about it. It feels like nobody is talking about how impossible and ridiculous all of this is - it’s one thing that OpenAI has promised one trillion dollars to people - it’s another that large swaths of that will be spent on hardware that will, by the end of these agreements, be half-obsolete and generating less revenue than ever. Think about it - let’s assume we live in a fantasy land where OpenAI is somehow able to pay Oracle $300 billion over 5 years. Said money is paying for access to Blackwell…
October 9, 2025 at 11:28 PM
Reposted by bicboss
bicboss.bsky.social
lmao apparently sore 2 already got the enshittification treatment, that was fast
October 6, 2025 at 10:40 AM
Reposted by bicboss
edzitron.com
Hey I talk and think about this all the time. This circular “just asking questions” routine is a classic of the genre for AI boosters. They do not want to come to an agreement with you, they want you to roll around in the mud until you doubt yourself. Do not waste your time
shrikant.bsky.social Shrikant Joshi @shrikant.bsky.social · Oct 5
This assumes that LLM responses are random which is incorrect. LLM responses are clearly context-based to the question asked.

Can you do the same to dice?

Then again, even dice are constrained by the number of faces they have. You can't get a 20 on a d6 or d12, can you?
October 5, 2025 at 11:46 AM
Reposted by bicboss
edzitron.com
They frame themselves as victims as they draw you into the equivalent of a 4am debate with a stoned teenager saying “but, like, what IS a mind?” And frame fair-minded skepticism as somehow restrictive because you can’t debate their endless, banal and directionless counterpoints. Walk away!
October 5, 2025 at 11:50 AM
Reposted by bicboss
edzitron.com
Honestly this is so cool. Finally the sheen is coming off this damn company. A great week for OpenAI - releasing the most expensive model of all time that also commits copyright violations as a service, you spent $9.2bn on compute in six months, and you can’t make a product better than the rabbit r1
October 5, 2025 at 4:55 AM
Reposted by bicboss
edzitron.com
A woman saying “The Al bubble is 17 times the size of the dot-com frenzy - and four times subprime, this analyst argues” and a man opening a calendar to see Ed Zitron the smiling man
October 3, 2025 at 3:39 PM
Reposted by bicboss
edzitron.com
This is another 10,000 word opus - I break down the entirety of OpenAI's finances, and explain that they're just another boring AI startup trying to build products on top of their models.

Sora 2 is a mirage - an unsustainable marketing tool.

Anyway, subscribe at the discount link below.
OpenAI Is A Boring AI Startup Making Novelty Products That Lose Billions Of Dollars One argument against this piece is that Sora itself represents a kind of innovation, when the truth is that it’s the result of infrastructure and theft. While there is some innovation with regards to post-training and the actual model itself, OpenAI’s moat is that they have the largest amount of available AI compute in the world and the resources to train massive models. They also intend to release the Sora 2 API, though it’s unclear how much it’ll cost (other than the fact it’ll be very, very expensive) and who might use it. In any case, OpenAI is kind of proving my point. Sora 2 is free - it doesn’t drive any revenue unless you consider the knock-on effects of people paying to access it, except you don’t get access to Sora 2 by paying for ChatGPT Plus, and still have to wait in line like everybody else. While theoretically opening up signups might increase the revenue driven by this product, it is still fundamentally a very expensive viral marketing tool, and every user it attracts is a chance to burn whatever revenue they may or may not bring in. And that, fundamentally, is the problem OpenAI faces. Even with a “viral hit” (if you can call it that) like Sora 2, one that’s got the entire tech industry talking, they still haven’t found a way to make a product people like or will actually pay for. The Sora 2 app is a clunky, ugly feed of slop, its generations unreliable and expensive, interesting not based on value or utility, but in the novelty of making something that sort-of-kind-of looks real, not that you can actually do anything with it as a result. OpenAI, just like its API customers, is stuck trying to work out how to make products that people want and will pay for, let alone enough to cover their horrifying costs. Sora 2 is legally-questionable, financially-reckless, and a terrifying act of desperation by a company that already spent at least $2.60 to make $1 in the…
October 3, 2025 at 1:16 AM
Reposted by bicboss
edzitron.com
Premium tomorrow: OpenAI spent $11.2 billion in compute, sales/marketing for $4.3bn in revenue in 1H 2025.

Desperate to grow ChatGPT, they launched Sora 2, which costs them at least $5 per generated video.

Here's $10 off a year of annual.
edzitronswheresyouredatghostio.outpost.pub/public/promo...
Sora 2 Is Unbelievably Expensive, With Each Video Costing $5 Of Compute and Users Allowed To Generate 100 Videos Every 24 Hours - Meaning OpenAI Will Lose A Month’s Subscription Revenue Every Four Videos Generated Let’s establish some facts: Once you have access to Sora 2, you can generate what appears to be an unlimited amount of videos, at least on a $20-a-month subscription, and OpenAI has promised “generous limits” for a non-specific amount of time. Every Sora 2 generation - in landscape or portrait - generates at 1280x704 - a variant of 720p resolution, and every single one is 10 seconds long, even if you ask it to generate a longer one. Each video appears to have sound, whether or not you asked for it. It takes about a minute to generate each video. Now, I’m a curious little critter, so I found that Microsoft actually publishes the prices of using Sora on its Azure AI Foundry page if you select “US East 2” as your region. To be clear, this is the pricing for the first Sora model, rather than Sora 2, which is likely expensive. Nevertheless, based on Azure’s pricing, it’s likely that each Sora 2 generation is priced at the “720 Global” rate of 50 cents a second, meaning that each video generated on Sora 2 costs $5 in compute, if not more, because it’s very likely that OpenAI’s costs are more than they charge on an API basis. Now, according to The Information, OpenAI’s gross margins were projected to be 50% this year, meaning that even in the most kindly, sweet maths of all time, they’re annihilating all of their margin in two generations, and another subscriber’s margins in two more. I also understand that some may argue that OpenAI’s costs “might not be as bad,” and there’s no evidence that this is the case. Sora 2, by OpenAI’s own description, is a “video and audio generation model,” suggesting it’s one main model rather than a combination with their audio-generation models they released earlier this year.
October 3, 2025 at 12:23 AM
Reposted by bicboss
edzitron.com
Playing around with Sora 2 and it is the same warmed up dogshit OpenAI always slings. I also got it to cough up direct copies of Naruto and Demon Slayer *with audio from the shows*. It’s obvious they have ingested a ton of copyrighted material
October 2, 2025 at 3:13 AM
Reposted by bicboss
edzitron.com
Here's me two months ago estimating - based on their revenue leaks - that OpenAI was at $5.26bn or so by end of July. I can't believe A) how close I was and B) how I was *too generous*. There were some who said I was being too harsh!
bsky.app/profile/edzi...
edzitron.com Ed Zitron @edzitron.com · Aug 1
OpenAI needs to make $12.7 billion by the end of the year, yet only has made around $5.26bn. To do this, they will have to make over $7 billion in four months, which will massively and aggressively increase costs.
www.wheresyoured.at/howmuchmoney/
Even If OpenAI Pulls It Off, Costs Are Astronomical (And Likely Higher Than Reported) Genuinely, OpenAI would have done something truly remarkable. It will have become one of the highest revenue-generating software companies in history, and have done so in record time. It will be genuinely impressive, though not super impressive, because it’s going to lose tens of billions of dollars doing so. The Information reports that OpenAI has "increased its cash burn to roughly $8 billion in 2025," but I see major inconsistencies in this number based on its own reporting. Last year OpenAI burned $5 billion on $3.6 billion or so of revenue. Compute costs are expected to be around $13 billion for Azure alone — nearly triple the $6 billion of last year, and that's before you consider CoreWeave and Google. OpenAI has more staff, needs more data, their hosting costs and administrative costs and research and development and every other god damn cost will likely be higher. Staffing, especially, is going to be a huge driver of its cash burn, and I’m not simply referring to its increased headcount. As mentioned earlier, in October of 2024, OpenAI had around 3,726 employees, who collectively received $1.5bn in stock-based compensation in the first half of that year (this figure does not include salaries). LeadIQ suggests its headcount is over 6000 today. We’re also in the midst of an AI hiring war, with researchers negotiating packages worth as much as $250m, according to the New York Times. The pressure to meet these demands will only increase the strain on OpenAI’s cash burn. How does that equal "burning $8 billion" exactly? If anything, I expect its burn to be more like $10 billion to $15 billion. Perhaps even more, depending on OpenAI’s hiring, salary, inflation, whether we count stock-based compensation as part of its burn (which, on accounting level, appears on a company’s profit and loss statement, even if it doesn’t involve cash transactions), and what costs the company… its burn (which, on accounting level, appears on a company’s profit and loss statement, even if it doesn’t involve cash transactions), and what costs the company has to pay as part of its Stargate Norway initiatives, as well as its other international expansion plans. That's the thing with OpenAI — even if it was to hit these unbelievable revenue targets again and again, it’s still doing so with a gaping wound in the side of its company, because the underlying economics of generative AI are so deeply unprofitable, with no sign of improving. Nevertheless, if OpenAI does this, it'll be genuinely interesting to see what happens next, because to hit its projected $26 billion in revenue in 2026, they will likely have to make more than Netflix (around $3 billion a month in revenue) by the end of the year.
September 30, 2025 at 6:19 PM
Reposted by bicboss
matthewhughes.bsky.social
My latest newsletter is about how big tech doesn't care about you.

Big tech will make you homeless, force you to drink poisoned water, and even kill you if it benefited them.

These are not good people.

whatwelost.substack.com/p/they-want-...
They Want To Ruin Your Life
It's Time To Accept That Big Tech Hates You
whatwelost.substack.com
September 30, 2025 at 4:05 PM
Reposted by bicboss
leyawn.bsky.social
meme of a man next to a stack of dominos going from very small to very large text over the largest domino: “openai needs one trillion dollars” text over the smallest domino: “ancient sumerian inventing the concept of money”
September 26, 2025 at 3:16 AM
Reposted by bicboss
edzitron.com
Tomorrow's premium newsletter: on analysis of their partnerships and projections, I estimate that OpenAI needs ~$1 trillion over the next 4 years, with at least $500bn needed for company operations.

Here's a $10 off discount off annual.
edzitronswheresyouredatghostio.outpost.pub/public/promo...
Compute Costs - At least $456.9 Billion Over Five Years Let’s run through this one more time: OpenAI has committed to spending $300 billion on compute through 2030 with Oracle. OpenAI has said it plans to spend an additional $100 billion on backup compute over the next four years. It’s unclear whether this figure includes the commitments stated previously, or are something distinct. Based on the wording of the coverage I’ve read, I’m inclined to think it’s a separate commitment — which is not good! OpenAI has agreed to buy $11.9 billion, $4 billion and $6.5bn in compute from CoreWeave over the next five years. Per The Information, OpenAI expects to spend at least $13 billion in 2027 and $28 billion in 2028 on compute with Microsoft. This also matches up with The Information’s reporting — though I actually think it’s going to be higher, as OpenAI fudges its numbers and “rounds” them (see the diagram). OpenAI Will Need To Raise At Least $500 Billion Just To Fund Its Operations, And Another $432.5 Billion Or More To Fulfil Its Obligations - Which Is More Than The Available Capital Of The Top 10 Private Equity Firms and US Venture Capital Combined So, I am repeating myself a little, but I really, really do not trust OpenAI’s revenue projections, and think they are at best unrealistic or, at worst, a deliberate attempt to mislead investors using the media. Nevertheless, even if OpenAI is in a position where it is making $200 billion by the year 2030, it will need to have raised at least $500 Billion to get there, and when you add its obligations — including those given to NVIDIA — it will have to raise (or have partners raise) another $400 or so billion. This is more than the projected available supply of US venture capital at the end of 2025 ($164 billion), and the combined available capital of the top 10 private equity funds ($477 billion), as well as the $50 billion of available direct lending capital from JPMorgan and Goldman Sachs (about $35 billion across two closes).
September 25, 2025 at 9:34 PM
Reposted by bicboss
raycarrot.bsky.social
Yeah so after hearing about the person behind the company producing this my excitement instantly turned to disappointment.

This could have been such a cool thing. Why did Ubisoft have to partner with these guys 😔 I'm not buying this anymore as I don't want to support someone like this.
raycarrot.bsky.social RayCarrot @raycarrot.bsky.social · Sep 23
This is so cool 👀

Kinda pricey though. I would have hoped they'd improve the game a bit, like make the Ubi Key level accessible by default and having actual save files instead of a password save, but I'm still getting this for sure!
raymangame.bsky.social Rayman (unofficial) @raymangame.bsky.social · Sep 23
Our friends at @modretro.bsky.social are releasing a special Chromatic x Rayman bundle! Check it out over at modretro.com/products/rayman.

To celebrate, we're doing a giveaway!
September 23, 2025 at 5:32 PM
Reposted by bicboss
hideokojima.bsky.kojimaproductions.jp
September 23, 2025 at 10:16 AM
Reposted by bicboss
edzitron.com
A woman saying “The Al boom is unsustainable unless tech spending goes 'parabolic, Deutsche Bank warns: "This is highly unlikely'” and a man opening a calendar seeing the smiling man ed Zitron
September 23, 2025 at 1:07 PM
Reposted by bicboss
edzitron.com
This was fantastic. We say “touch grass” like a curse on someone, when really the digital realm has become a place to separate us from the sensual (and I’d argue the emotional) connection we have with the world around us and those closest to us
What emerges is something like an upside down Gnosticism, a world designed to remove you from the sensual, and thereby rob you of community, autonomy, or self-knowledge. As Audre Lorde writes in her essential essay Uses of the Erotic: "When we live outside ourselves, and by that I mean on external directives only rather than from our internal knowledge and needs, when we live away from those erotic guides from within ourselves, then our lives are limited by external and alien forms, and we conform to the needs of a structure that is not based on human need, let alone an individual's...in touch with the erotic, I become less willing to accept powerlessness, or those other supplied states of being which are not native to me, such as resignation, despair, self-effacement, depression, self-denial." What people want to replace you with is what they think you already are. If companies, bosses, and boyfriends want to replace you with a robot or an AI model, that's because they already see you and what you do as simply the workings of a machine with far too many bugs. When we buy into this framing, we lose half the battle. The way out is through the senses. Reading before bed, eating without a screen in front of you, spending less time scrolling, trying to go for walks without your phone, if you have tried and failed at any or all of the above, it might be because you are simply denying yourself of something designed in a lab to give you dopamine, and are not replacing it with something sensuous. You need to lean back into the feelings of your body to escape the dehumanizing technology of the wealthy and their devices of control. Why do we touch grass? To touch grass. In a way, I think there might be a doorway to something beautiful that is created in all the rejection, frustration, and hatred being felt right now towards things like AI, social media, the Cracker Barrel logo thing, and the general enshittification of the internet. It is important to not just hate this stuff, but to ask ourselves why we hate it. I think what we may find is that deep under all that anger, is a love of what is human. The messiness, the flaws, the seeming imperfections of being a body and not a machine.
September 22, 2025 at 3:42 PM
Reposted by bicboss
edzitron.com
Me in December 2024 referencing my work from earlier in 2024

www.wheresyoured.at/godot-isnt-m...
Throughout these pieces I have repeatedly made the point that — separate to any lack of a core value proposition, training data drought, or unsustainable economics — generative AI is a dead end due to the limitations of probabilistic models that hallucinate, where they authoritatively state things that aren't true. The hallucination problem is one that is nowhere closer to being solved — and, at least with the current technology — may never go away, and it makes it a non-starter for a great many business tasks, where you need a high level of reliability.
September 21, 2025 at 4:16 PM
Reposted by bicboss
edzitron.com
A woman with a speech bubble says “OpenAI admits AI hallucinations are mathematically inevitable, not just engineering flaws”. A man opens a calendar to see ed Zitron, the smiling man
September 21, 2025 at 4:27 PM
Reposted by bicboss
qwazor.co.uk
trans rights are human rights 🏳️‍⚧️
September 19, 2025 at 9:23 AM
Reposted by bicboss
katmabu.bsky.social
Trans rights are human rights.
September 17, 2025 at 2:54 PM
bicboss.bsky.social
80 dollars port collection
September 12, 2025 at 2:27 PM
Reposted by bicboss
matthewhughes.bsky.social
My latest newsletter is about how the double standards that allows big tech to commit the most appalling crimes -- and to get away with them.

whatwelost.substack.com/p/big-tech-a...
Big Tech Always Escapes Justice
How Google, OpenAI, and others get away with (literal and figurative) murder
whatwelost.substack.com
September 4, 2025 at 6:39 PM