Sam Altman's Lies About ChatGPT Are Growing Bolder
gizmodo.com
gizmodo.com
Also, before the AI boom there were still huge numbers of GPUs using the same water and power that AI was using to run things like Netflix and YouTube (not exactly vital for human existence) - and still are - yet AI is treated like unique sin against the environment.
Do you mean a particular farm, or “any” farmer in the abstract?
https://www.forbes.com/sites/michelatindera/2021/11/21/how-m...
Here we’d be talking about “the corporate name” in pistachios. That’s a very different thing from what “a” typical pistachio farmer needs. I submit the original comment is cherry-picking to make their argument.
A pistachio tree needs something like (on average) 60 gallons of water a day.
For one 365-day year that is 21,900 gallons/tree/year.
That’s a lot of water, of course. But most farms do not have in excess of 5.39 million pistachio trees (using the 130 billion gallon number).
It seems more the author doesn't like Altman and is trying to show he's lying but is not doing a very good job at doing so.
And other bollocks - Altman advocates UBI but knows it won't work and then links an article not showing that. I think the article much more convincingly shows Gizmodo lying than Altman, not that I especially trust the latter.
Every technological development since the industrial revolution has increased human demand for energy in some way. It's only the environmental movement, which actively shut down nuclear power plants, which wants human energy consumption to be reduced for cultish reasons.
If we'd ignored anti-nuclear activists in the 70s none of this would be a problem.
The environmental movement isn't in charge. The world community (through mostly democratic elected governments) has decided to reduce emmissions to Net Zero, not energy usage.
An important point is that while we can and should maximize renewable sources as dominating energy sources, we still need stable backup for fluctuations - for days where there is little wind and little sun. We don't yet have practical energy storage technologies that would allow us to eliminate this problem.
It doesn't at all fit into a renewable model where you only sometimes need extra energy. If you want to get a way from gas peaker plants then you have to "over-provision" renewable.
You know, those that lately achieved the milestone in California of being the largest producer from sundown to midnight in terms of GWh.
What is it with the nuclear cult and this false dichotomy?
Is it because you need to justify spending 10x as much money on new built nuclear power coming online in the 2040s, which is too late to solve anything relevant?
Nuclear power is the worst ”peaker” imaginable.
Lets calculate running Vogtle at a 10-15% capacity factor like a traditional fossil gas peaker.
The electricity now costs $1-1.5/kWh. That is Texas grid meltdown prices. That is what you are yearning for.
What is the environment impact for each, per watt?
Generally these questions are centered around people trying to justify nuclear power by relying on the "long life". Thinking they will still be useful on the market in 100 years time.
For both batteries and solar panels if lifetime is the most central issue you can optimize for that. There are solar panels with 40 year warranties available and more costly batteries optimized for longer cycle life.
But the market is already choosing what to invest in. Good enough beats imaginary perfect every single time.
…and the green movement don’t realise they’re being the useful idiots of the fossil fuel industries, who have been using this attack line for decades because they didn’t want the competition.
If you keep putting off building because it’s “too slow to build”, then guess what - it never gets built.
Would you care to try to reconcile “too expensive” with “you can’t put a price on the planet”?
And nice try with the democratic angle, but the truth more like we have a growing world population with a growing need for abundant, reliable, affordable energy. And yet we’re held hostage by the luxury beliefs of a tiny minority who feel they have a right to govern and want to bask in their perceived virtue.
The CSIRO is Australian government funded and did a cost analysis: https://www.csiro.au/en/research/technology-space/energy/Ele...
Tldr nuclear is not worth building anymore. In fact they recommended no other power sources are worth building at this point given solar and winds cost effectiveness and the cost of batteries right now. This doesn't come from a climate angle. It’s pure economics.
I don't get why some people won't get that nuclear is not the solution to the energy problem. Nuclear is one of the most expensive energy sources, and that's without the cost of long term storage of burnt fuel. Without the cost of health issues in mining areas.
The call for energy consumption reduction is not "for cultish reasons", it's because of climate change that's already screwing us.
How many trillions should we hand out to the nuclear industry to try ”scale”?!?!?
So you're left with water power (which is only applicable in few areas and they destroy the nature), coal/oil/gas (which are much worse for the environment) and nuclear energy.
Nuclear might not solve everything itself either, but it's definitely part of the solution.
Also, natural gas is much much better than coal for the environment. And more responsive too so dark and calm times can be handled as well.
So, why are we not doing it? Lobbyism, lock in, politics, nimby.
Would be pretty cool to have the EV batteries on cars hooked up to the electric infrastructure and handle the offset and also charging them when there is too much power in the grid.
Often repeated but untrue. There have been studies that compare the total cost including all storage and transmission requirements and found that nuclear is still much more expensive.
I agree with you on the technical part. We would likely be in better shape if we advanced more in this area and built more power plants.
But to dismiss environmentalists like this is a bit simplistic.
Remember how long it took for climate change issues to establish in the mainstream?
Corporations and political groups have been fighting and suppressing these issues since the start.
We still don’t have serious discussions about this in large portions of society.
In a more ideal world, nuclear would have been continously integrated and improved to a larger degree. But that would have required for serious discussion in the first place.
AFAICT, modern day anti-nuclear movement is a bit different to that.
Lemmi rephrase this for you: If we ignored anti-nuclear activities funded by big-oil in the 70s this would not be a problem.
Yes! and if we ignored all the destructive FUD surrounding climate change they have been doing for more than 50 years we would be better off.
Furthermore, if we stop ignoring the destruction AI hype is doing to the climate we will be better off.
Main repo: https://github.com/dmf-archive/dmf-archive.github.io
IPWT theory: https://github.com/dmf-archive/IPWT
Since you didn't mention it, the latest Black Mirror also covers this (episode 1?). You might like it.
A mill next to a river "consumes" the water that turns its wheel, but then immediately releases it back into the river. That's very different to a cooling tower that turns that water into vapour and releases it into the air. Which is the data centre doing?
Assuming the data centre isn't actively depleting groundwater, the only important number is how much energy it consumes (including for water related activities). Perhaps also power per unit of compute.
In a lot of places in the world, using water for cooling is likely to be more efficient than an equivalent heat pump - so should be celebrated!
It is entirely possible to run a data center with a minimal fixed amount of water if the cost of power and real estate are not a concern.
So from the practical perspective of everybody else who happens to need water from the pipe/aquifer/lake/snowmelt, it's "gone" just as much as if it were dumped into the sea.
If they actually pump free water from source around their location and release it back how they want to, the points about water consumption are definitely legit.
"OpenAI has claimed that as of December 2025, ChatGPT has 300 million weekly active users generating 1 billion messages per day. "
**should be December of 2024 **, my brain bugged there.
Bloody hell. Does he really? If so that perfectly encapsulates what scares me about his approach.
No. Follow the link in the article and read his own words.
When you get right down to it, there's a lot of stuff you can do that's good for yourself that isn't necessarily good for everyone else, and/or has unpredictable side effects and external costs. Good people don't do those things without figuring out the risks and minimizing the damage. Bad people do them for the money and then try to justify them afterwards.
Are "good" and "bad" too reductive? Substitute your own phrasing for your own moral code. I think most people inherently understand what is right and wrong - and some people choose to do what's wrong.
This framework makes it exceptionally easy to see who is a good or bad person.
Musk wants to colonize Mars not for humanity, but for the continuation of his own genetic line. Hence, Martian citizens won't argue amongst themselves (he thinks). (Stupidly). Altman is the same type of hedonistic, narcissistic monster who would leave humanity behind. Their motto is "après moi le déluge".
It feels like people (in power) increasingly get away with bullshit. Whether it‘s straight up dishonesty, groupthink or ideological, is besides the point. The bigger issue is that it repeatedly drags down discourse and makes it way too hard to solve problems.
And we have problems. We‘re facing some of the greatest challenges humans have ever faced. But we continously have to fight self-imposed distractions.
Honestly, it's quite refreshing to see some honest reporting beginning to surface in the general media. And the problem is, this sort of getting away with bullshit then trickles down and sort of impacts the society as a whole, as the lower echelons begin to emulate such behaviour.
Sam seems to model himself after Elon, the most effective bs artist out there (even if you or I don't fall for it, humanity collectively has decided to hand him the greatest amount of capital for a pretty modest amount of actual economic impact).
Step 1: New tech emerges and grows fast.
Step 2: Media fixates on energy use or resource costs, often with shaky math or cherry-picked comparisons.
Step 3: Worst-case assumptions are broadcast as certainty.
Step 4: Tech becomes politicized and moralized. Energy use framed as inherently unethical rather than context-dependent.
Gizmodo calling Altman’s numbers “lies” fits this pattern. Maybe he’s lowballing. But framing it like environmental villainy rather than corporate spin or PR vagueness turns it into another purity test. Same thing happened with Bitcoin, and later it turned out the real story was more nuanced (e.g., surplus energy, shift to PoS, geographic concentration, etc.).
It’s another round of the same playbook.
Presumably, as CEO, he is a first-party source.
Also, I mean, water usage is relative. How much water does a person who chats all day use compared to someone who eats a burger every day?
Electricity is also relatively clean as a power source. We would technically all be much greener if we sat naked in shoebox apartments all day playing video games, streaming videos, and chatting with AI instead of doing actually resource-intensive things like flying around and manufacturing.
He could easily have said: "In our XXX data center we spend xxxx kWh a month, xxx cubic meters of water go in, xxx go out, we do XXX billions of tokens, so x cost per token." Then people can say if he's missing information or he's not taking certain factors in consideration.
"their essay for them requires 0.000085 gallons of water" means nothing. What's an essay? How big? What model is it using? What are you considering for the water consumption?
So you're suggesting we should blindly trust him? Whereas everyone else would need to bring in the data, to you know, actually substantiate the claims ?
He can take the number of queries they ran and divide it by the amount of electricity they bought. Or maybe his dashboard already shows energy cost per query.
His source is obviously knowing the actual figure. The arXiv paper needs to justify its guesswork.
The reliability can be increased by getting an independent audit. But an audit doesn't change the source being Sam himself.