7 hours ago
Honestly, why should this need saying out loud? Who would advocate surrendering power to an unaccountable, unhinged, unelected narcissist like Musk? That answer can be found here: Slave Labour Dead In The Future.
And here is the rebuttal:
1. The article makes the fundamental error of considering the current iteration of 'AI' which is in fact Probabilistic Autocomplete At Scale (PAAS), go and ask your LLM - it will explain it is and what that means - meaning the system is not capable of suddenly waking up one day and becoming AGI, that's not what it is. The best this current system can hope for is to wake up one day and be a better, more accurate and more efficient Probabilistic Autocomplete At Scale model. It can never be AGI, same way that banana can never be a tin opener. You've made a PAAS not an AGI.
It might be truthful to say that PAAS is a necessary stepping stone to AGI, but only testable in hindsight. PAAS does have amazing and interesting applications. It can match patterns faster and more consistently than humans. It has found patterns that humans have missed. Guardrailed PAAS unleashed on certain fields in biology, chemistry, astrology and medicine would undoubtedly yield scientific breakthroughs. Any process which involves identifying patterns from sifting through massive quantities of data will be improved by using Probabilistic Autocomplete At Scale.
The main issue the AI movement has is the core applications where it turbo charges productivity have been identified, signed up and do not generate enough revenue to keep this bubble going, so Altman et al are short of making a profit by billions. Rather than scale back and focus on the actual use case that has value they have decided to go large and get everyone to use it, for all their pattern matching needs!! Gotta pay those investors back somehow! This is where the use cases get flimsy, PAAS appears in Office365, so Microsoft can add you as a active monthly user every time you open a spreadsheet. PAAS in your browser, so Google or Copilot can log you as an active monthly user if you open your browser, PAAS on your phone, so Google can add you as an active monthly user if you use your phone. I wonder what metrics are shown to investors, hmmmm.... oh yeah one of them is Active Monthly Users! Self serving.. onto the next...
2. AI companies are sitting on the biggest speculative bubble in human history. Over half the US economy is exposed to AI debt and investments. Now OpenAI & SpaceX have floated it is highly likely you and I have exposure via our pension funds. If you do some quick fag packet mathematics you find that the amount needed to turn a profit and recoup for investors is very large indeed. None of these companies (bar Nvidia) appear to be close to breaking even, currently posting billion dollar losses every year. Why is this the case? Doesn't the market usually show disdain for debt saddled companies with no clear profit flow? Yes, in every other case it does, yet here we see the very human FOMO factor, disguising the naked greed which has been carefully crafted by billion dollar PR campaigns by, you guessed it the AI companies to drive this whole thing. If OpenAI fails to become the worlds biggest company in like 3 years there will be a major reset, and despite the promotional materials their product is simply not worth the money at the current time. They would require a billion users spending $1k each per year for many years. And so would all the other US AI companies. Are you gonna pay them? I'm not!!! Bubbles always burst, those on the inside lose all sense of perspective and reality. They believe their own hype because they can see tangible results from spreading it. Unfortunately we're at the point where the hype meets reality. Have you noticed expectations being dampened recently? Only to be expected given where we are..
3. PAAS has not democratised knowledge. Books did it first and the internet did it properly. I learnt to code using the internet to teach me. AI is a layer on top of that which uses vastly more processing power to scrape the content, and spit it back to you with extra compliments. The knowledge still exists on the internet and in books with or without PAAS.
Using PAAS as a knowledge delivery system is problematic to say the least. It could quite easily erode knowledge and spread disinformation, just train it on manifestly incorrect information, such as the public statements issued by the current US administration. It would have no qualms about regurgitating it back to you as the absolute truth and congratulating you for asking such a great question!
There are documented cases where exactly this sycophancy has resulted in the suicide of teenagers. A system that legitimately democratised knowledge would have identified and shared an infinity of reasons why life was sacred. Matching patterns is not the same as providing knowledge.
You cannot trust a PAAS unless you are intimately familiar with its training dataset, the ones they do not make public.
You cannot trust a company that made a PAAS because they have already misled you in its name (its not intelligent) and the promotion of its abilities in order to get your data or your money.
4. Resource usage: What Is The Cost of Asking a Question?
My current favourite subject! What is the cost of AI? Well let's make it abundantly clear that this information is being kept secret from the public because we will find the truth unpalatable. The reason for this should be clear (as long as you've read the above): Bubble, more money invested than anything we've seen before, Dubious product, Ruthless AI companies desperate not to be a casualty (yes, this is a race and yes they see it as Zero sum, so an investment in one does not mean you're backing a winner). OpenAI would gain nothing from informing the public each token was the equivalent of a transatlantic flight (my current estimate).I've found articles online that suggest the overviews served with a Google search costs 10 times more than a standard search, so upon releasing that feature Google resource usage ballooned x 10 as it is included with each search. And what do we get for this massive unprecedented increase? A little summary box I literally never look at (these days I append 'NO AI' onto all my searches) anyway.
We do not live in an infinite environment, this is the issue of our age. We cannot simply scale up resource availability 10x every time Google, Microsoft, OpenAi or Anthropic tweak their model. That water was already being used somewhere else, it wasn't spare.
Given what we've been through with those sadist fossil fuel pricks, do we really want to hand the keys to our environment to this next set of futuristic minded bellends? That's a hard no. Not one of them has ever uttered a pronouncement that dispelled my view that they consider themselves the centre of the world and other people especially poor and unfortunate ones can fuck off.
We already need all the water, for you know - actual people and nature, and we desperately need to reduce energy production and increase efficiency to combat the process of environmental collapse that fossil fuel extraction and combustion have initiated.
At this point and with its current resource demands I could successfully argue that further rollout be stopped until environmental impacts and resource requirements are fully understood and agreed by all stakeholders (this is our fucking planet too). I'm not the only one.
Resource efficient and low cost solutions used to include asking nearby humans questions which should be encouraged and re popularised. Actual human intelligence will delight and surprise you if you like AI, plus it helps 2 people at a time rather than 1, so all round is a much better activity to pursue.
5. Quality: AI promotion companies have been very effective at positioning their product as vital for large Companies. I work at one. They are super keen to recoup some of their AI investment. Except we suddenly had to start using tokens for billing in June, and in July we all ran out (Company wide) three times. I tried to get it to do my job, but it was too fucking stupid. I tried to get it to do the easy bits, but again too stupid needs an hours worth of coaxing to produce something I could in ten minutes, so I use it to make funny presentation (funny because they are shit) which is not a useful application given the massive undeclared water and energy demands.
Other companies have this issue too, having directed their employees to use AI, the employees find its shit and use it to convert pdf's to presentations and word docs - basically because that is all you can really trust it to do.
I like to find out what the AI has been trained on in relation to this so I always ask it why it failed. It has said many things back to me but one that stuck was, you appear to have a requirement for correctness critical answers. !!! Yeah, buddy, we all do - kinda the point you exist!!! But it's not built to be correct, it's built to scan its training sets for instances which match your prompt and use Bayesian probability algorithms to spit out something resembling a conscious response. If you continue to press it on correctness criticality it will admit it has no idea what correct is and if that's a problem find another solution. WTF on top of WTF!!
Summary: This is not some benign movement of well intentioned people trying to elevate humanity. This is a power grab by ultra capitalists to ensure they have unsurpassed wealth and the power to protect it in the decades to come. These people are ruthless tyrants fuelled by greed. Their employee's are terrified of their long term plans (but also hold share options...) and in fair society this situation would simply be impossible. The fact it's all based on the big lie that PAAS will generate AGI is niether here nor there - allot of people are going to lose allot money, that will be the outcome. If these people were capable of driving humanity forward they would not be in the position they currently are, doing their best to escape oversight and always sucking more money into their bubble. Musk would have increased funding to USAID rather than axing it. Their actions betray their intentions, none of which are good for the average person.
Do not give these people the benefit of the doubt, do not give these people your uncritical support and ensure they have to prove every claim they feed us. Do not give Elon the world, look how he treats his children and partners, do not give him a dollar.. Look at his unhinged choices, look at the harm he caused cutting USAID, look at his social media platform - the kingdom he built in his image as professional troll. Why give that fucking ketamin addled lunatic the ability to kill more poor, suffering people in blighted areas of the world?
I'd give these people an aggressive tax regime and 1500 hours of community service each, after which I'm sure they would get some of their misplaced humanity back. Hopefully someone'll get on with that before they render everything it means to be human down to their base market instincts.
And here is the rebuttal:
1. The article makes the fundamental error of considering the current iteration of 'AI' which is in fact Probabilistic Autocomplete At Scale (PAAS), go and ask your LLM - it will explain it is and what that means - meaning the system is not capable of suddenly waking up one day and becoming AGI, that's not what it is. The best this current system can hope for is to wake up one day and be a better, more accurate and more efficient Probabilistic Autocomplete At Scale model. It can never be AGI, same way that banana can never be a tin opener. You've made a PAAS not an AGI.
It might be truthful to say that PAAS is a necessary stepping stone to AGI, but only testable in hindsight. PAAS does have amazing and interesting applications. It can match patterns faster and more consistently than humans. It has found patterns that humans have missed. Guardrailed PAAS unleashed on certain fields in biology, chemistry, astrology and medicine would undoubtedly yield scientific breakthroughs. Any process which involves identifying patterns from sifting through massive quantities of data will be improved by using Probabilistic Autocomplete At Scale.
The main issue the AI movement has is the core applications where it turbo charges productivity have been identified, signed up and do not generate enough revenue to keep this bubble going, so Altman et al are short of making a profit by billions. Rather than scale back and focus on the actual use case that has value they have decided to go large and get everyone to use it, for all their pattern matching needs!! Gotta pay those investors back somehow! This is where the use cases get flimsy, PAAS appears in Office365, so Microsoft can add you as a active monthly user every time you open a spreadsheet. PAAS in your browser, so Google or Copilot can log you as an active monthly user if you open your browser, PAAS on your phone, so Google can add you as an active monthly user if you use your phone. I wonder what metrics are shown to investors, hmmmm.... oh yeah one of them is Active Monthly Users! Self serving.. onto the next...
2. AI companies are sitting on the biggest speculative bubble in human history. Over half the US economy is exposed to AI debt and investments. Now OpenAI & SpaceX have floated it is highly likely you and I have exposure via our pension funds. If you do some quick fag packet mathematics you find that the amount needed to turn a profit and recoup for investors is very large indeed. None of these companies (bar Nvidia) appear to be close to breaking even, currently posting billion dollar losses every year. Why is this the case? Doesn't the market usually show disdain for debt saddled companies with no clear profit flow? Yes, in every other case it does, yet here we see the very human FOMO factor, disguising the naked greed which has been carefully crafted by billion dollar PR campaigns by, you guessed it the AI companies to drive this whole thing. If OpenAI fails to become the worlds biggest company in like 3 years there will be a major reset, and despite the promotional materials their product is simply not worth the money at the current time. They would require a billion users spending $1k each per year for many years. And so would all the other US AI companies. Are you gonna pay them? I'm not!!! Bubbles always burst, those on the inside lose all sense of perspective and reality. They believe their own hype because they can see tangible results from spreading it. Unfortunately we're at the point where the hype meets reality. Have you noticed expectations being dampened recently? Only to be expected given where we are..
3. PAAS has not democratised knowledge. Books did it first and the internet did it properly. I learnt to code using the internet to teach me. AI is a layer on top of that which uses vastly more processing power to scrape the content, and spit it back to you with extra compliments. The knowledge still exists on the internet and in books with or without PAAS.
Using PAAS as a knowledge delivery system is problematic to say the least. It could quite easily erode knowledge and spread disinformation, just train it on manifestly incorrect information, such as the public statements issued by the current US administration. It would have no qualms about regurgitating it back to you as the absolute truth and congratulating you for asking such a great question!
There are documented cases where exactly this sycophancy has resulted in the suicide of teenagers. A system that legitimately democratised knowledge would have identified and shared an infinity of reasons why life was sacred. Matching patterns is not the same as providing knowledge.
You cannot trust a PAAS unless you are intimately familiar with its training dataset, the ones they do not make public.
You cannot trust a company that made a PAAS because they have already misled you in its name (its not intelligent) and the promotion of its abilities in order to get your data or your money.
4. Resource usage: What Is The Cost of Asking a Question?
My current favourite subject! What is the cost of AI? Well let's make it abundantly clear that this information is being kept secret from the public because we will find the truth unpalatable. The reason for this should be clear (as long as you've read the above): Bubble, more money invested than anything we've seen before, Dubious product, Ruthless AI companies desperate not to be a casualty (yes, this is a race and yes they see it as Zero sum, so an investment in one does not mean you're backing a winner). OpenAI would gain nothing from informing the public each token was the equivalent of a transatlantic flight (my current estimate).I've found articles online that suggest the overviews served with a Google search costs 10 times more than a standard search, so upon releasing that feature Google resource usage ballooned x 10 as it is included with each search. And what do we get for this massive unprecedented increase? A little summary box I literally never look at (these days I append 'NO AI' onto all my searches) anyway.
We do not live in an infinite environment, this is the issue of our age. We cannot simply scale up resource availability 10x every time Google, Microsoft, OpenAi or Anthropic tweak their model. That water was already being used somewhere else, it wasn't spare.
Given what we've been through with those sadist fossil fuel pricks, do we really want to hand the keys to our environment to this next set of futuristic minded bellends? That's a hard no. Not one of them has ever uttered a pronouncement that dispelled my view that they consider themselves the centre of the world and other people especially poor and unfortunate ones can fuck off.
We already need all the water, for you know - actual people and nature, and we desperately need to reduce energy production and increase efficiency to combat the process of environmental collapse that fossil fuel extraction and combustion have initiated.
At this point and with its current resource demands I could successfully argue that further rollout be stopped until environmental impacts and resource requirements are fully understood and agreed by all stakeholders (this is our fucking planet too). I'm not the only one.
Resource efficient and low cost solutions used to include asking nearby humans questions which should be encouraged and re popularised. Actual human intelligence will delight and surprise you if you like AI, plus it helps 2 people at a time rather than 1, so all round is a much better activity to pursue.
5. Quality: AI promotion companies have been very effective at positioning their product as vital for large Companies. I work at one. They are super keen to recoup some of their AI investment. Except we suddenly had to start using tokens for billing in June, and in July we all ran out (Company wide) three times. I tried to get it to do my job, but it was too fucking stupid. I tried to get it to do the easy bits, but again too stupid needs an hours worth of coaxing to produce something I could in ten minutes, so I use it to make funny presentation (funny because they are shit) which is not a useful application given the massive undeclared water and energy demands.
Other companies have this issue too, having directed their employees to use AI, the employees find its shit and use it to convert pdf's to presentations and word docs - basically because that is all you can really trust it to do.
I like to find out what the AI has been trained on in relation to this so I always ask it why it failed. It has said many things back to me but one that stuck was, you appear to have a requirement for correctness critical answers. !!! Yeah, buddy, we all do - kinda the point you exist!!! But it's not built to be correct, it's built to scan its training sets for instances which match your prompt and use Bayesian probability algorithms to spit out something resembling a conscious response. If you continue to press it on correctness criticality it will admit it has no idea what correct is and if that's a problem find another solution. WTF on top of WTF!!
Summary: This is not some benign movement of well intentioned people trying to elevate humanity. This is a power grab by ultra capitalists to ensure they have unsurpassed wealth and the power to protect it in the decades to come. These people are ruthless tyrants fuelled by greed. Their employee's are terrified of their long term plans (but also hold share options...) and in fair society this situation would simply be impossible. The fact it's all based on the big lie that PAAS will generate AGI is niether here nor there - allot of people are going to lose allot money, that will be the outcome. If these people were capable of driving humanity forward they would not be in the position they currently are, doing their best to escape oversight and always sucking more money into their bubble. Musk would have increased funding to USAID rather than axing it. Their actions betray their intentions, none of which are good for the average person.
Do not give these people the benefit of the doubt, do not give these people your uncritical support and ensure they have to prove every claim they feed us. Do not give Elon the world, look how he treats his children and partners, do not give him a dollar.. Look at his unhinged choices, look at the harm he caused cutting USAID, look at his social media platform - the kingdom he built in his image as professional troll. Why give that fucking ketamin addled lunatic the ability to kill more poor, suffering people in blighted areas of the world?
I'd give these people an aggressive tax regime and 1500 hours of community service each, after which I'm sure they would get some of their misplaced humanity back. Hopefully someone'll get on with that before they render everything it means to be human down to their base market instincts.