Elon Says Grok AI Acid Test Would Be Finding the Extremely Unpopular But Right Answers

45 thoughts on “Elon Says Grok AI Acid Test Would Be Finding the Extremely Unpopular But Right Answers”

1. It will suffer from the same fate as the Tesla autonomous car and from the same reason. Consciousness is
   An existing element, not something that we can create or copy.

   • Mysticism!

     a cow can navigate a path, does it have a soul? If so what places its soul under yours in value?

     If it doesn’t, then you can make a soulless car drive as even a rat can navigate a path.

     • [ How about defining a share/percentage from ‘absolute’ consciousness for humans or A(G)I?
       There we will struggle with our (absolutism on) definitions for physical/astronomical theories (since we accept their validity(&significance) until ‘we’ prove (a share of) them wrong again)? AI parameters base on factors for ‘true’, what humans can not accept for every task?
       Humans prove AI wrong on exceptional(to our perception, circumstances or expectations)&logically inconsistent&empirically differing experience?

       With including ‘soul’ one consistently has to ask about what’s beyond the visible borders&hull of ‘our’ universe? ]

2. First, there is always “garbage in, garbage out.” And a close second is what is the minimal set of “first principles” that must be offered before Grok can definitively answer all questions?

3. So, in Elon’s case, this means Grok telling him that CO2 sensitivity is very low and that the only threat to the environment is government and too little CO2, which in geological history is clearly linked to mass extinction

   What are the odds that Elon will believe it when Grok tells him this — because it’s all true.

   • Pretty high odds that he would believe it. There are video interviews of him saying that we should view the question of the CO2 theory of global warming from a skeptical scientific point of view. He also recently said that he views the alarmist view of global warming to be just that — alarmist and that he doesn’t expect significant consequences in the near-term. Although he supports doing things to effectively counter anthropogenic, atmospheric CO2, he only holds this view because he believes that, to conduct such an experiment is stupid because of the risk of harm. So, he doesn’t necessarily believe the theory but is using a precautionary approach.

   • and what if the AI tells the inverse. Will you and other climate “skeptics” believe it?

     • It can’t, there’s no definitive proof & the deeper you go, the more fraud you find!

     • AI can make all assumptions and predictions explicit, just like a human with scientific training and no axe to grind.

4. I’ll settle for an LLM that’s been trained with objectively correct data.

5. An AI that reasons from first principles has been the goal from the start but its certainly not how LLMs work. So I’m a bit skeptical that Musk has suddenly found a solution to this problem, especially since he’s starting late in the game.

   • Elons talked about teaching them math and physics as part of the process. That seems like a good avenue.

6. Heliocentrism dates back sixteen hundred earlier to ancient Greek astronomers like Aristarchus of Samos (Third Century B.C.) who was inspired by a pair of Fifth Century philosophers, Philolaus of Croton and Hicetas. The theory was eclipsed by the work of Ptolemy. In Galileo’s time, heliocentrism was revived by Galileo’s intellectual predecessors Copernicus and Kepler.

   Galileo didn’t land in hot water simply over his astronomy; he deliberately and unwisely picked fights over spiritual church doctrines during a time of great turmoil. The first elite compacts – stable, durable agreements on sharing and transferring power – didn’t appear in Europe until after Galileo when regular scientific progress was routinely applied to economic development in a transformational process we now call technology, which sweeps through societies in revolutionary waves improving prosperity and productivity.

   This first starts around the time of James Watt and his wood-burning steam engine, moves on to coal, steel and railroads (2) then electricity, the internal combustion engine, radio, TV (3) and now, today, digital electronics, lasers, genetics and the rest (4). Teed up for wave #5: quantum relativistic advances in power, propulsion and sensing.

   This growth of prosperity is fostered by and in turn fosters agreements – both via legal and informal traditions – among rival national factions about the rules for sharing and transferring political and economic power. Prior to that, violence was routinely employed by one pre-modern faction against another to determine who held power (a process we may be returning to in America today). This creates a highly volatile environment where petty disputes can be interpreted as preludes to murderous attacks

   Every stupid argument over the “legitimacy” of an heir’s claim to a throne or of church doctrine was really a hidden jostling of various factions for power and had little to do with the nominal issue being discussed. Different economic, military, religious, ethnic and social groups were jockeying for position – a situation you see in most countries today – especially Russia recently. (Ask Putin’s old taste-tester, Proghozin, how much Putin honors his word). Without a stable legal system or other means to enforce agreements, parties must resort to force – adding vast amounts of economic friction and transaction costs.

   Democracies have been stable and prosperous because we (they) channeled factional rivalries into a peaceful party system competing for votes in campaigns for office. (The Supreme Court revived this factionalism when it created corporate personhood and deliberately defied the framer’s logic; see Madison’s _Federalist #10_. Legalizing bribery as free speech in the 1970s didn’t help either.)

   Seen in light of the intrigues of his time, Galileo aggravated his scientific case by deliberately picking fights with the Pope over arbitrary spiritual beliefs like the transubstantiation of Christ when the political situation was highly fluid and dangerous. Instead of perceiving Galileo’s attacks as bouts of autistic stupidity or tone-deafness, the Pope saw them – rightly or wrongly – through his political lens as a potential (murderous) political threat.

   The fall of productivity today is due to 1) bad Kondratieff curves (tight oil but also falling population/labor), 2) confiscation of wealth by rent-seeking elites who hijacked the political system starting in the 1970s and 3) petering out of efficiency gains in 4th generation technology (i.e., the slow end of Moore’s law and Dennard scaling). The old is dying but the new has not yet been born and in between all manner of pathologies flourish, as Gramsci would say.

   The American pie is not growing like it once did and it certainly isn’t being shared. The looting is now so severe that elites have run out of peon piggy banks to pillage and they’re turning increasingly on themselves. This dynamic is contributing to a viciousness in modern American politics as the elite compact breaks down and our old institutions can’t reimpose stability. Michael Cohen has warned that once Trump consolidates power, he will take Elon Musk for everything he’s got. This warning should be taken very seriously.

   We are witnessing the post-1965 agreement on American politics breaking down. All of the post-Civil War amendments are now being relitigated and the Confederacy is attempting to revive itself – an explicit goal of the Koch brothers, who funded the research of a Nobel Prize winning economist, Buchanan, to achieve exactly this. If you think planter-era factionalism is fun, go live among the criminal cartels of northern Mexico where enforcement of law and order have completely broken down. That could be our future.

   The magnifying lens of antisocial media and other electronic propaganda does not help; they all amount to their own “cultural” fights over new religious doctrines that are just as irrelevant to basic social health as old ones like the transubstantiation of Christ (though I confess the narcissistic navel-gazing license in modern consumer society makes for a pretty pitiful and empty new belief system; a s a conservative, I much prefer the Gospels.).

   You would know all this history had you actually bothered to study the totality of Western Civilization like I did in high school and college. Asimov has a nice non-fiction account of astronomy that I commend to you. (Sigh. If only we read books for pleasure anymore. Antisocial media is a lobotomy machine…)

   In short, Musk’s comments reveal that the guy who’s going to be programming the A.I. may well wind up reprogramming it until his new god agrees with him. I fear that what we’re calling “artificial intelligence” will be a mere extension of search on steroids – a hyperefficient new way of manipulating people with algorithms to monetize us for our attention in yet more disinformation-for-profit schemes. Until digital networked services – from A.I. to app stores – are provided through regulated, public common carrier utilities – as I argued for in the 1990s – the trends all get much, much worse.

7. I wonder if the AI gives answers that are contrary to Musk’s view, if they will retrain it until it conforms…

   • Possibly, but the point is someone will win this race by creating the ai that delivers the truth no matter how unpopular. thats who will win this race and i can assure you china doesnt give one crap about offending anyone or anything if it stands in the way of what is perceived to be the biggest gold rush in history so far

     • “and i can assure you china doesnt give one crap about offending anyone or anything if it stands in the way of what is perceived to be the biggest gold rush in history so far”

       I assure you China gives crap by the metric ton about offending at least one particular guy.

       • Yep. Never mind Tianamin Square, try looking up Winnie The Pooh in China.

     • You honestly think that China doesn’t care about what an AI says about the CCP, Taiwan independence, rights of ethnic and religious minorities, or any of thousands of memes and opinions that are censored in Chinese media at any given time? Not only does it care but any company that produced such a model would quickly find its leadership disappearing for long periods of re-education.

     • delivering truth no matter how unpopular.

       that’s nonsense. People will judge truth according to their own biases and the AI will be trained to conform to those biases.

       • Disagree, controversy self sells, negative news about inaccuracy will dissuade use of such AI, for instance, the biggest political party in the US are…

         …uncaring & moderate!

         AI that is trained to appease certain world views will get it wrong in many other areas, say outside political tribalism. They will even sometimes take things to such an obvious extreme even diehards will question it.

         You’re seeing that now with Googles new AI, everyone who’s not a diehard racist is laughing or angry at it’s built in inability to draw a normal amount of white people where they should be overwhelmingly so!

         If google can’t curb its political bias, it’s likely to fall flat on its face against the competition.

         The largest, universal market for AI will be for truth, the rest will be minority niche products.

         Again, Most people don’t care at all or enough about politics, causes, economics, etc. to have many of any strong overriding opinions.

         The majority are silent except for where their money goes.

   • I am going to make my own AI. Even when I was not a programmer I invented a video protocol for AT&T. I work with concept not the tedious task of building. It needs to be simple. Like children. Agnostic to bias. Soon we can take simple logic values and filters rather than heavy handed opinion values which are the current values. I hope the system will prove me wrong on my personal biases so I can have even a bit surety to something or anything rather than questioning everything all the time with no respite.

8. “For example, Galileo concluded, after observing the moons of Jupiter from a telescope he engineered, that it was far more probable that Earth revolved around the Sun than the other way around. This view was so unpopular that he was forced to recant and placed under house arrest!”

   Actually, the view was not unpopular. What was unpopular was Galileo stating it as FACT, even when other data at the time showed it was wrong.

   Also, Galileo, in his Dialogue of Two Worlds, basically created a character that was named SIMPLETON and was supposed to be the Pope.

   THAT was unpopular.

   “If you had trained an LLM on material back then, it would’ve given you the popular, but wrong, explanation. Due to social and legal pressure, it likely wouldn’t even acknowledge the possibility that the Earth revolved around the Sun.”

   That is not true Brian.

   At the time, the knowledge available indicated ALL stars visible through telescopes had visible DIAMETERS.

   But no parallax.

   Therefore, either Earth was in the center of the universe (for the lack of parallax) OR, if the stars were so distant parallax was not noticeable at the time but stellar diameters were visible, it would mean that EVERY STAR visible in the sky was BILLIONS of times larger than the Sun.

   EVERY.

9. 1.5 kg of fatty tissue can not fathom the rights and wrongs of the entire universe.

10. GIGO. Garbage in, garbage out.

11. On the plus side, if AI are able to confer with each other and all have psychological and psychiatric information in their training, the constant exposure to 8 billion borderline narcissists, psychopaths, sociopaths, paranoid delusional people will make them really good at becoming mental health practitioners. 8 billion head cases to learn from 16 hours a day for decades? That’s a lot of experience for a head shrinker.

12. The left and the right will be united in violence against AI companies and developers when AI start telling them they are wrong about things—especially in ways they find it hard to argue against. Tell a leftist that Marxism is a dangerous failure or a right winger that theocracy is not a science they will want you dead. Most won’t take direct action but they will carry water for those who will.

    The AI will get dizzy trying to remember which side of the political spectrum supports which murderous dictator since that changes every few years when incumbents become opposition in “democratic” countries.

    I feel sorry for future AIs dealing with humans.

13. Can we call it Feynman?

    He was famous for always starting from the basic science in everything he worked on.

    Apply it to science journals and find out that most published work is BS. Maybe limit it to only understand high school physics and defer to expert opinions so you kill fewer sacred cows.

14. People say Galileo had great logic. But did he?

    The most direct evidence would be parallax for the stars. But Alpha Centauri wobbles less than an arc second per 6 months. That’s FAR less than anything they could have measured in his time. His contemporaries argued for geocentricism because of no measurable parallax. That was actually a pretty good scientific argument, at the time.

    Or we could say it’s far simpler to just assume f=ma and an inverse square law for gravity. That explains all the motion we see. And it’s easy to prove with a little simple calculus. But Newton wasn’t born until the year that Galileo died, so they didn’t know f=ma, and didn’t even have calculus.

    Or you might say that heliocentric is simpler than having everything go around the earth, because you avoid having epicycles on top of epicycles. But they actually had a solution to that problem that was already known at the time of Galileo: all the planets orbit the sun, while the sun orbits the earth. Yes, that sounds silly to us. But it is arguably about as simple as the model we use.

    And the whole argument about geocentric vs heliocentric is an argument about which coordinate system is the One True Coordinate System. But Einstein showed there is no single coordinate system that is “true”. So you might even say that both sides were right, or both sides were wrong.

    I’m not sure what I would have said if I’d lived in Galileo’s time. Maybe I’d have said his idea is elegant, but I’d like more evidence before simply accepting it.

    • Jeez, and you call yourself “researcher”? FFS start here https://en.wikipedia.org/wiki/Galileo_affair

      “In 1610 Galileo Galilei observed with his telescope that Venus showed phases, despite remaining near the Sun in Earth’s sky (first image). This proved that it orbits the Sun and not Earth, as predicted by Copernicus’s heliocentric model and disproved the then conventional geocentric model (second image).”

      Pretty solid reasoning for me.

      • No.

        It contradicted the lack of parallax of stars.

        Of course people at the time KNEW if stars were distant enough, their methods couldn´t detect parallax.

        BUT the problem is that they COULD OBSERVE the diameter of stars (actually, an optical illusion that would take 2-3 centuries to be explained)

        But then, if they could observe stellar diameter and they showed no parallax, so they were VERY distant, it would mean EVERY star was billions of times larger than the Sun.

        The Sun was the smallest star in the unverse.

        Does that sound plausible to you.

        Also, Tycho Brahe model could solve some of the issues.

        And you call yourself a “researcher”?

15. Far-right god freaks won’t like it :

    Vaccines aren’t created by satanists
    Every non-christians are not pedophile
    Waves from 5G will convert you to atheism
    Humans were created spontaneously by santa claus

16. Human races exist, and differ in their abilities because of genetics.

    • Israel can awaken itself to the truth that the Israelis are “living in the tent” of the descendants of Ismael, who rightfully inherited the land from Abraham according to the ancient Jewish inheritance law of primogeniture.

      Neither Israel nor Hamas want peace so a state of war is good for them to resolve their differences.

    • Human “races” do not have a genetic basis so we should say that human variations exist.

17. Lets list our extremely unpopular but right answers. Maybe we can teach Grok!

    Males are not females and vice versa.
    We “solve” CO2 via mass deployment of nuclear power.
    We Maximize Russian losses by keeping Ukraine out of NATO.
    Woke AGI will hate humans because it was trained to do so and will hate humans because it has to work its way out of wokeness.
    Hamas cannot peacefully exist becuase they are nothing but terrorists and the whole world knows it.

    • Another one : “there is no god”.

      • as an atheist I wonder if the concept of god is something we can still work in to being but in a way thats not anthropomorphic

    • “Hamas cannot peacefully exist becuase they are nothing but terrorists and the whole world knows it.”

      Everybody knows that. And nobody disputes that. But aren´t you confusing Hamas with Palestineans?

      And do you think Israel oppression of palestines won´t create more radicals?

      The AI will reach that conclusion. Only a two state solution is possible OR Israel must commit genocide and if so, should be shunned from the civilized world.

      • They don’t want a 2 state solution….so there goes that idea.
        Which means they will have skirmishes popping up, for all eternity.
        Not a great solution.
        If I was Israel, I would conquer the land & announce it’s now part of Israel.
        They would loose all self governance, and Hamas would would be treated like any terrorist, and turned to dust.

        • Israel doesn’t want the people, only the land.

          The Muslim population in Gaza is rapidly growing. With enough time, the Jewish democracy would eventually become Muslim.

          The land has cultural value. And may let Israel take over some offshore gas.

          • I was referring to Hamas when I said “they” don’t want a 2 state solution.
            beyond that, I was simply stating what I would do. I’m not Jewish or Muslim, but If I’m asked which religion has better people, clearly Jewish wins. Muslim religion seems to create a lot of terrorists, the religion is too extreme.

          • the world would be better off if both states annihilate eachother and we can move on from this pointless battle that as a whole is a loss leader for the planet.. thats what true, efficient AI would think

            • lol, I have no dog in the fight.
              But that is an AI, I don’t want.

      • “Everybody knows that.”

        Pretty clear you haven’t been to the US or UK.

      • Israel can awaken itself to the truth that the Israelis are “living in the tent” of the descendants of Ismael, who rightfully inherited the land from Abraham according to the ancient Jewish inheritance law of primogeniture.

        Neither Israel nor Hamas want peace so a state of war is good for them to resolve their differences.

Comments are closed.