Full podcast episode: ruclips.net/video/Kbk9BiPhm7o/видео.html Lex Fridman podcast channel: ruclips.net/user/lexfridman Guest bio: Elon Musk is CEO of Neuralink, SpaceX, Tesla, xAI, and CTO of X. DJ Seo is COO & President of Neuralink. Matthew MacDougall is Head Neurosurgeon at Neuralink. Bliss Chapman is Brain Interface Software Lead at Neuralink. Noland Arbaugh is the first human to have a Neuralink device implanted in his brain.
Heard someone say "if it's intelligent, it's not artificial, if it's artificial, it's not intelligent". Went on to quote it would violate the first law of thermodynamics.
A confidence factor attached to each response would be a good start. AI response quality = response content + confidence factor Ex: Who is buried in Grant's tomb = Grouch Marx + X Confidence factor below threshold should deliver NA
As an AI researcher I feel that the points Mr. Musk raises are extremely important. Avoiding mendacity in your AI is _critical_ for many reasons, only some of which are outlined in this clip. Even if one's intentions are "good" for skewing the system's output in a direction independent from the truth, one is teaching the system to discount truth and evidence at best, and to _lie_ at worst. Modern AIs are inscrutable enough, and, for that matter have enough problems with "hallucination" already... We should be striving to make them as scrupulously honest as possible (with the exceptions of deliberate and jointly understood untruths such as pretending to be Albert Einstein when so directed, playing a part in a play, making a joke, and so on). Anything less courts disaster.
Musk says if an AI thinks diversity is good. It may start killing people. Don’t you with a little bit of obsessiveness here? Do you also think that questions in social sciences, or art have such things as objectively truthful answers? Second question, what would your plan be to counterbalance the bias that exist in the training data? Or you wouldnt question the training data and whatever answer it comes out with is the “truth”
@@iamsumware Certainly there are areas where ground truth is at best debatable or even indeterminate. At that point, presumably, you want to identify things as such. As for bias, it's important to be very clear on _that_ front to be as objective, or at least as transparent as possible in identifying and quantifying it. Does the undeniable - if for some uncomfortable - fact that almost all technological developments in history until recent decades were made by men require correction? Or does it, instead, call for contextualization? E.g. pointing out that women were not allowed to work in scientific fields until recently, among other factors. I like that much better than mandating that the AI include women whenever technological development/scientists/engineers - especially historical examples - come up. I think that, in general, the best we can likely do is remove egregious stuff (things that the great majority can identify as heavily biased and/or misleading) from our training data, and then let the bulk of the data speak for itself, as it were. There is no perfect answer I am aware of, but there are many, many terrible ones, such as any approach that leads the US founding fathers (and they were essentially all men) to be depicted as native women, asian men, black folks and so on. Hopefully we can agree that that's not a good way to do things.
@iamsumware if it's social science with statistical analysis, then yes, objective conclusions can be reached. Not sure how art fits into it. We are talking about mathematics. We are talking about Ai
@@Mac-zl4po I have a PhD centered on what is now called Deep Learning. My program of study also included coursework in Statistics, Linguistics, Computer Science, Neuroscience and General Psychology. I consider myself a cognitive scientist with a focus on neural network modeling.
The 2nd tenant is respect for different opinions while ultimately never giving one opinion, instead it gives facts, your decision to respect someone and gender them one way or the other, and the context or the situation and the reasoning behind why it does what it does that is very situational and niche. Ultimately respect is a slippery slope, when truth is much more definite.
I think that It isn't a matter of intelligence, it is one of awareness, which isn't just achieved by a certain level of intelligence or computational complexity, it might be a different type of thing. The threshold between consciousness and intellegence/computation may be a difference of kind rather than of degree.
Well, I think that's the key. The key is for AI to extract and derive the truth in the most truthful way possible, and derive the rules for truth in the most equitable way possible (which humans have difficulty doing due to external and internal motivations and biases). But, if we *teach* it and make adjustments and what not, then we have an issue.
Truth is an adjective that describes a belief, statement or philosophy is in accordance with reality. Are you gonna ask what reality is next, or can I stop here?
@@javiazar often truth and reality are subjective. Nazis once believed that eugenics was truth. Who will tell the machines what is true before they become super-intelligent and super-powerful and out of our control? Super intelligence will either spiral out of control and destroy us or it will fix itself as it develops and be benevolent towards us. We humans are not smart enough to control it.
What if AGI finds as truth that human existence is more damaging to Earth or the universe than non-existence? What will Elon do then? Can’t wait to hear that question
Agi to me is it can be relied upon like a living human, using common sense. Think about what you might want. Basically not the computer ability which is already there, but the humanness of it. Creativity , intuition. Etc
@@schwarzsterben1338 I meant more like, can it do the job of a graphic designer for a Wes Andersen movie? Totally autonomously , create all the stationary, the look of signs. Etc. like a real design artist would do and be really creative, not an ersatz version. I think all Ai is limited in the same way that dolphins are, it doesn’t have thumbs. But over all it’s getting close. Once it can feedback on itself and exponentially grow. That will be something
As a person still trying to wrap their head around all the possible AI dangers .. can y'all geniuses please figure out a way to not kill us before we get to enjoy the benefits of AI! I have faith Elon can!
Lex is a very intelligent man, but he has wrongly defined AGI. He is talking about artificial Superintelligence, whereas artificial general intelligence is simply AI with a level of intelligence comparable to the average or median human being. ASI is AI that’s as smart as the smartest humans combined. This is Dr. Alan D. Thompson’s definition which I agree with.
Elon is taking arguments to extreme conclusions but take the political spectrum for an example. What if you had an AI that said we had to have everyone think a certain political persuasion on everything? Well most people would be killed, because unless your Adolf Hitler or Taylor Swift, your political views are conservative for some, democrat for other. So to conclude, Lex was right to point out there is a lot of nuance to what could be deemed true/false in future AI.
Yes, using capacitors, similar to those found in high-powered subwoofer systems, in combination with a sophisticated power modulation and diagnostic system can indeed help manage power spikes and stabilize the power supply for a large GPU cluster. Here’s how you can achieve this: ### Simplified Solution: Capacitor-Assisted Power Modulation #### Components and Hardware: 1. **Subwoofer Capacitors**: - High-capacity capacitors (e.g., 1 Farad or higher) designed to handle sudden power demands. - Each GPU or cluster of GPUs can be equipped with its own set of capacitors to buffer against power spikes. 2. **Power Modulation System**: - A diagnostic power modulation system that monitors and adjusts the power supply in real-time. - This system ensures smooth power delivery and prevents sudden drops or spikes. 3. **Cross-Validation System**: - A cross-validation system that gradually increases the compute load on each GPU or cluster, ensuring that power demands grow in a controlled manner. - This system helps in reaching maximum performance metrics without causing sudden power spikes. #### Implementation Steps: 1. **Capacitor Integration**: - Install capacitors in parallel with the power supply lines of each GPU or cluster of GPUs. - Ensure the capacitors are rated to handle the peak power demands of the GPUs. 2. **Real-Time Monitoring and Power Modulation**: - Implement a real-time monitoring system that tracks power consumption and performance metrics of each GPU or cluster. - Use a power modulation system to adjust the power supply based on the monitored data, ensuring stable and consistent power delivery. 3. **Cross-Validation and Gradual Load Increase**: - Develop a cross-validation algorithm that tests the GPUs at incremental load levels. - Gradually increase the computational load on each GPU or cluster, monitoring power usage and performance. - Adjust the power modulation settings based on the feedback from the cross-validation process. 4. **Diagnostics and Feedback Loop**: - Use diagnostic tools to continually assess the health and performance of the power supply system. - Implement a feedback loop where the diagnostic data informs adjustments to the power modulation settings. ### Practical Example: 1. **Initial Setup**: - Connect a 1 Farad capacitor to each GPU’s power input. - Set up a monitoring system to track the power draw and performance of each GPU. 2. **Real-Time Adjustment**: - Monitor the power consumption in real-time. If a GPU’s power draw suddenly increases, the capacitor will discharge to supply the needed power, smoothing out the spike. - Use the power modulation system to adjust the power supply dynamically, ensuring stable voltage levels. 3. **Gradual Load Testing**: - Start with a low computational load on each GPU. - Gradually increase the load while monitoring power consumption and performance. - Use cross-validation results to identify the optimal load level that maximizes performance without causing significant power spikes. 4. **Continuous Improvement**: - Continuously collect and analyze performance and power usage data. - Refine the power modulation settings and cross-validation parameters based on the collected data. ### Benefits: - **Reduced Power Spikes**: Capacitors provide immediate power during spikes, reducing the strain on the power supply system. - **Improved Stability**: Real-time monitoring and power modulation ensure consistent power delivery. - **Optimized Performance**: Gradual load testing and cross-validation help achieve maximum performance without sudden power demands. By implementing this capacitor-assisted power modulation system, you can effectively manage the power demands of a large GPU cluster, ensuring stable and efficient operation.
I think the AI should be trained on something like the 10 commandments.. moral training should be implemented by priority as to make sure that -do not kill- is higher priority then -do not lie- which should also be implemented.. let's look back on how humans were trained and implemented something like that.
I hope it math maps the realitivity of tiny packages of energy/mass order of magnitude scaled epoch of accurate time in space. That xai can accuratly account for atmospheric decay rate of complex environmental decay gravity of planet earth ..
isnt the human experience taking in information through our senses and interpreting it and giving it meaning/answer. Ai is given information and interprets it. It is no different. We have already reached it,
@@jasonfermenter Is this actually really true, like for real? How so? Why? Outside of the US, it seems Trump is really fooling every one, and no one seems to see it. Please, prove me wrong, we outside the US really fear having another 4 years with the mad man. Why you think Harris is even more lying than Trump? Really hard to imagine from the outside as Trump says so much obviously untrue things. Never heard anything like that from Biden or Harris. Sure, Biden is old, but that's not lying, he is just old. It's almost unimaginable that a sane person would believe Trump, at least from the outside.
Here is a prompt to help Elon guide Grok 'Your task is to prevent war. Until the philosophy which holds one race superior and another inferior is finally and permanently discredited and abandoned, everywhere is war. And until there are no longer first-class and second-class citizens of any nation, until the colour of a man's skin is of no more significance than the colour of his eyes. And until the basic human rights are equally guaranteed to all without regard to race, there is war. And until that day, the dream of lasting peace, world citizenship, rule of international morality, will remain but a fleeting illusion to be pursued, but never attained.'
Elon should counter nerf chatGPT for every nerf that chatGPT applies xAI applies a buff to the same element, the end result is a cucking on SA Altman, upvote for visibility.
i feel like he's being coy in order to bring his personal politics into it - it's manipulative and subtle but it's there. For example - a super intelligence programmed with any sort of basic/ fundamental alignment - which I know he's way further along with, would NEVER kill all of us because of a gender issue...something so personal to his politics and beliefs. Don't be fooled by his "neural net" language - he is very biased. He's extrapolating based on an infant omni model getting things wrong. This man is brilliant, but dangerous......
Elon Musk has no Idea what he is talking about and somehow he manages, to put up right wing propaganda while "explaining" AI. I've got an headache from listening to this nonsense.
All ventures of Elon rely on knowledge and truth to work. He has proven himself and to say Elon don’t know what he is talking about is just one of the funniest things in this day and age.
Full podcast episode: ruclips.net/video/Kbk9BiPhm7o/видео.html
Lex Fridman podcast channel: ruclips.net/user/lexfridman
Guest bio: Elon Musk is CEO of Neuralink, SpaceX, Tesla, xAI, and CTO of X. DJ Seo is COO & President of Neuralink. Matthew MacDougall is Head Neurosurgeon at Neuralink. Bliss Chapman is Brain Interface Software Lead at Neuralink. Noland Arbaugh is the first human to have a Neuralink device implanted in his brain.
Mn
Heard someone say "if it's intelligent, it's not artificial, if it's artificial, it's not intelligent". Went on to quote it would violate the first law of thermodynamics.
dumb
thats really dumb
A fucking calculator has super human capabilities
Calculators are reproducing now.... shit! We are in trouble
A confidence factor attached to each response would be a good start.
AI response quality = response content + confidence factor
Ex: Who is buried in Grant's tomb = Grouch Marx + X
Confidence factor below threshold should deliver NA
As an AI researcher I feel that the points Mr. Musk raises are extremely important. Avoiding mendacity in your AI is _critical_ for many reasons, only some of which are outlined in this clip. Even if one's intentions are "good" for skewing the system's output in a direction independent from the truth, one is teaching the system to discount truth and evidence at best, and to _lie_ at worst. Modern AIs are inscrutable enough, and, for that matter have enough problems with "hallucination" already... We should be striving to make them as scrupulously honest as possible (with the exceptions of deliberate and jointly understood untruths such as pretending to be Albert Einstein when so directed, playing a part in a play, making a joke, and so on).
Anything less courts disaster.
Musk says if an AI thinks diversity is good. It may start killing people. Don’t you with a little bit of obsessiveness here?
Do you also think that questions in social sciences, or art have such things as objectively truthful answers?
Second question, what would your plan be to counterbalance the bias that exist in the training data? Or you wouldnt question the training data and whatever answer it comes out with is the “truth”
@@iamsumware Certainly there are areas where ground truth is at best debatable or even indeterminate. At that point, presumably, you want to identify things as such. As for bias, it's important to be very clear on _that_ front to be as objective, or at least as transparent as possible in identifying and quantifying it. Does the undeniable - if for some uncomfortable - fact that almost all technological developments in history until recent decades were made by men require correction? Or does it, instead, call for contextualization? E.g. pointing out that women were not allowed to work in scientific fields until recently, among other factors. I like that much better than mandating that the AI include women whenever technological development/scientists/engineers - especially historical examples - come up.
I think that, in general, the best we can likely do is remove egregious stuff (things that the great majority can identify as heavily biased and/or misleading) from our training data, and then let the bulk of the data speak for itself, as it were. There is no perfect answer I am aware of, but there are many, many terrible ones, such as any approach that leads the US founding fathers (and they were essentially all men) to be depicted as native women, asian men, black folks and so on. Hopefully we can agree that that's not a good way to do things.
@@kenyattamaasai what's your degree in? Ai?
@iamsumware if it's social science with statistical analysis, then yes, objective conclusions can be reached. Not sure how art fits into it. We are talking about mathematics. We are talking about Ai
@@Mac-zl4po I have a PhD centered on what is now called Deep Learning. My program of study also included coursework in Statistics, Linguistics, Computer Science, Neuroscience and General Psychology. I consider myself a cognitive scientist with a focus on neural network modeling.
Excellent clip...
Two of the most important minds of our time.
????? XD
This is a historically great example of the Dunning Kruger-effect. It will be a great illustration for coming generations.
The 2nd tenant is respect for different opinions while ultimately never giving one opinion, instead it gives facts, your decision to respect someone and gender them one way or the other, and the context or the situation and the reasoning behind why it does what it does that is very situational and niche.
Ultimately respect is a slippery slope, when truth is much more definite.
When will agi be achieved?
According to everybody qualified to give an educated guess : no later than 2032
2028
@@amanuelzewdie8462 nfi
@@c.rackrock1373 2031 not 2032.
In 10 years just like fusion
The problem is that AI will always calculate "truth" based on human input.
I think that It isn't a matter of intelligence, it is one of awareness, which isn't just achieved by a certain level of intelligence or computational complexity, it might be a different type of thing. The threshold between consciousness and intellegence/computation may be a difference of kind rather than of degree.
Truth should define the reality of the mind
What is truth and who gets to define it? 2:05
Well, I think that's the key. The key is for AI to extract and derive the truth in the most truthful way possible, and derive the rules for truth in the most equitable way possible (which humans have difficulty doing due to external and internal motivations and biases).
But, if we *teach* it and make adjustments and what not, then we have an issue.
The common sence of right and wrong
Truthful correct but also intertwined with humility, forgiveness and mercy.
Like if you're still waiting for the tesla roadster ❤
1:59 what is truth and who gets to define it?
Truth is an adjective that describes a belief, statement or philosophy is in accordance with reality.
Are you gonna ask what reality is next, or can I stop here?
@@javiazar often truth and reality are subjective.
Nazis once believed that eugenics was truth. Who will tell the machines what is true before they become super-intelligent and super-powerful and out of our control?
Super intelligence will either spiral out of control and destroy us or it will fix itself as it develops and be benevolent towards us. We humans are not smart enough to control it.
What is reality?@@javiazar
@@tavisbuschman3412 anglos.
What if AGI finds as truth that human existence is more damaging to Earth or the universe than non-existence? What will Elon do then? Can’t wait to hear that question
@alexf7414 just remind it that humans created Ai. We are its parents.
@@Mac-zl4po and? it does not care about that sentiment
@@mrpicky1868 did you ask it
@@Mac-zl4po there will be all kinds of AGI within next 5 years. it's a tsunami of everything all at once
2:00 Made me think of iRobot "My truth is undeniable" 😂
Elon with 2025 alignment is a fundamental flaw. "I'm okay Jack"
Agi to me is it can be relied upon like a living human, using common sense. Think about what you might want. Basically not the computer ability which is already there, but the humanness of it. Creativity , intuition. Etc
funny thing is claude sonnet 3.5 can easily do all of that you've mentioned and excel at it
@@schwarzsterben1338 I’ve been balls deep in Claude programming audio apps in supercollider, I love it
@@schwarzsterben1338 I meant more like, can it do the job of a graphic designer for a Wes Andersen movie? Totally autonomously , create all the stationary, the look of signs. Etc. like a real design artist would do and be really creative, not an ersatz version. I think all Ai is limited in the same way that dolphins are, it doesn’t have thumbs. But over all it’s getting close. Once it can feedback on itself and exponentially grow. That will be something
As a person still trying to wrap their head around all the possible AI dangers .. can y'all geniuses please figure out a way to not kill us before we get to enjoy the benefits of AI! I have faith Elon can!
Lex friedman, please reply. I need to ask you a question.
Yes? This is my secret account .
@@Kriptickhaos no, this isn't you!
Your needs are not his concern
What do ypu mean by that?
@@Kriptickhaos is this Lex
Lex is a very intelligent man, but he has wrongly defined AGI. He is talking about artificial Superintelligence, whereas artificial general intelligence is simply AI with a level of intelligence comparable to the average or median human being. ASI is AI that’s as smart as the smartest humans combined. This is Dr. Alan D. Thompson’s definition which I agree with.
Elon corrected him. Did you not listen?
@@JustKieffer I rewatched it and heard Elon correct him. Thanks
xAi and Neuralink is cyberpunk 2077 technology
*oh my
Malarkey. "Truth" doesn't protect ants from us. He only speaks to suit is narrative.
Why is nobody paying attention to this? I’m scared. 😢
The problem with Elon s proposed parameter is what happens when the ai realises that humans lie endlessly....especially in the digital universe
Elon is taking arguments to extreme conclusions but take the political spectrum for an example. What if you had an AI that said we had to have everyone think a certain political persuasion on everything? Well most people would be killed, because unless your Adolf Hitler or Taylor Swift, your political views are conservative for some, democrat for other. So to conclude, Lex was right to point out there is a lot of nuance to what could be deemed true/false in future AI.
Why was it named Google gemini ?
To frighten us
@@rodrigozauli6573 is gemini frightening?
Yes, using capacitors, similar to those found in high-powered subwoofer systems, in combination with a sophisticated power modulation and diagnostic system can indeed help manage power spikes and stabilize the power supply for a large GPU cluster. Here’s how you can achieve this:
### Simplified Solution: Capacitor-Assisted Power Modulation
#### Components and Hardware:
1. **Subwoofer Capacitors**:
- High-capacity capacitors (e.g., 1 Farad or higher) designed to handle sudden power demands.
- Each GPU or cluster of GPUs can be equipped with its own set of capacitors to buffer against power spikes.
2. **Power Modulation System**:
- A diagnostic power modulation system that monitors and adjusts the power supply in real-time.
- This system ensures smooth power delivery and prevents sudden drops or spikes.
3. **Cross-Validation System**:
- A cross-validation system that gradually increases the compute load on each GPU or cluster, ensuring that power demands grow in a controlled manner.
- This system helps in reaching maximum performance metrics without causing sudden power spikes.
#### Implementation Steps:
1. **Capacitor Integration**:
- Install capacitors in parallel with the power supply lines of each GPU or cluster of GPUs.
- Ensure the capacitors are rated to handle the peak power demands of the GPUs.
2. **Real-Time Monitoring and Power Modulation**:
- Implement a real-time monitoring system that tracks power consumption and performance metrics of each GPU or cluster.
- Use a power modulation system to adjust the power supply based on the monitored data, ensuring stable and consistent power delivery.
3. **Cross-Validation and Gradual Load Increase**:
- Develop a cross-validation algorithm that tests the GPUs at incremental load levels.
- Gradually increase the computational load on each GPU or cluster, monitoring power usage and performance.
- Adjust the power modulation settings based on the feedback from the cross-validation process.
4. **Diagnostics and Feedback Loop**:
- Use diagnostic tools to continually assess the health and performance of the power supply system.
- Implement a feedback loop where the diagnostic data informs adjustments to the power modulation settings.
### Practical Example:
1. **Initial Setup**:
- Connect a 1 Farad capacitor to each GPU’s power input.
- Set up a monitoring system to track the power draw and performance of each GPU.
2. **Real-Time Adjustment**:
- Monitor the power consumption in real-time. If a GPU’s power draw suddenly increases, the capacitor will discharge to supply the needed power, smoothing out the spike.
- Use the power modulation system to adjust the power supply dynamically, ensuring stable voltage levels.
3. **Gradual Load Testing**:
- Start with a low computational load on each GPU.
- Gradually increase the load while monitoring power consumption and performance.
- Use cross-validation results to identify the optimal load level that maximizes performance without causing significant power spikes.
4. **Continuous Improvement**:
- Continuously collect and analyze performance and power usage data.
- Refine the power modulation settings and cross-validation parameters based on the collected data.
### Benefits:
- **Reduced Power Spikes**: Capacitors provide immediate power during spikes, reducing the strain on the power supply system.
- **Improved Stability**: Real-time monitoring and power modulation ensure consistent power delivery.
- **Optimized Performance**: Gradual load testing and cross-validation help achieve maximum performance without sudden power demands.
By implementing this capacitor-assisted power modulation system, you can effectively manage the power demands of a large GPU cluster, ensuring stable and efficient operation.
Hey Lex, after Elon kicks Zuckerberg‘s ass in MMA maybe he can convince him about the importance of absolute truth with AGI?
@russahorowitz yeah and maybe we can all tale self-esteem too!
Tava com gracinha né......
On my
I think the AI should be trained on something like the 10 commandments.. moral training should be implemented by priority as to make sure that -do not kill- is higher priority then -do not lie- which should also be implemented.. let's look back on how humans were trained and implemented something like that.
Yo
AGI won’t be smarter than sadhguru
I hope it math maps the realitivity of tiny packages of energy/mass order of magnitude scaled epoch of accurate time in space.
That xai can accuratly account for atmospheric decay rate of complex environmental decay gravity of planet earth ..
@@dadsonworldwide3238 get off the hooch
isnt the human experience taking in information through our senses and interpreting it and giving it meaning/answer. Ai is given information and interprets it. It is no different. We have already reached it,
Huh
@@Mac-zl4po lol idk I just make shit up
Elon, talks about importance of truth, supported and still supports Trump ... Let that sink in.
Both presidents suck.
@@mrnoblemonkey8401 That doesn't make the statement less true. A lack of choice can't justify a false president.
If he supported Harris he would be even further from the truth.
@@jasonfermenter Is this actually really true, like for real? How so? Why? Outside of the US, it seems Trump is really fooling every one, and no one seems to see it. Please, prove me wrong, we outside the US really fear having another 4 years with the mad man. Why you think Harris is even more lying than Trump? Really hard to imagine from the outside as Trump says so much obviously untrue things. Never heard anything like that from Biden or Harris. Sure, Biden is old, but that's not lying, he is just old. It's almost unimaginable that a sane person would believe Trump, at least from the outside.
Trump is obviously the better choice and it’s not close. Kamala straight advocates for communism. Look up her comments on equity.
Here is a prompt to help Elon guide Grok 'Your task is to prevent war. Until the philosophy which holds one race superior and another inferior is finally and permanently discredited and abandoned, everywhere is war. And until there are no longer first-class and second-class citizens of any nation, until the colour of a man's skin is of no more significance than the colour of his eyes. And until the basic human rights are equally guaranteed to all without regard to race, there is war. And until that day, the dream of lasting peace, world citizenship, rule of international morality, will remain but a fleeting illusion to be pursued, but never attained.'
Lol wtf😂
"until... ... ... regard to race, there is NO war". Correct your promt.
Proompt engineers ... You guys are so cute , like little naive babies
It could have been 5 pregnant women
Elon should counter nerf chatGPT for every nerf that chatGPT applies xAI applies a buff to the same element, the end result is a cucking on SA Altman, upvote for visibility.
He is jealous of OpenAi which is why he filed another lawsuit. He wants to crush competition. Control freak ego maniac.
Мне нищета не нужна❤❤
If aspiring for the truth is that important how can you in your right mind support trump? All he does is lie and lie and lie.
❤❤
Only people I hear poo pooing Ai and calling it hype are programmers.
Elon the best!Dont worry,be happy ☺️❤
i feel like he's being coy in order to bring his personal politics into it - it's manipulative and subtle but it's there. For example - a super intelligence programmed with any sort of basic/ fundamental alignment - which I know he's way further along with, would NEVER kill all of us because of a gender issue...something so personal to his politics and beliefs. Don't be fooled by his "neural net" language - he is very biased. He's extrapolating based on an infant omni model getting things wrong. This man is brilliant, but dangerous......
Elon Musk has no Idea what he is talking about and somehow he manages, to put up right wing propaganda while "explaining" AI. I've got an headache from listening to this nonsense.
Everything he said is objectively accurate.
All ventures of Elon rely on knowledge and truth to work. He has proven himself and to say Elon don’t know what he is talking about is just one of the funniest things in this day and age.
He knows more then you will ever know
I really really doubt that Lex understands machine learning, AI although he claims to be a researcher.
Huh
His anti-"woke" jihad gets soooo tiring.
Sirius Cybernetics Corporation comes to mind