back to article Microsoft will force its 'superintelligence' to be a 'humanist' and play nice with people

Microsoft has joined the ranks of tech giants chasing superintelligent artificial intelligence, but the company's AI chief Mustafa Suleyman’s vision is markedly different from that articulated by other industry leaders To be clear, “superintelligent AI” hasn't been invented yet. Pundits use the term to describe systems that …

  1. Anonymous Coward
    Anonymous Coward

    Microsoft promises to pray to the Omnissiah that it's vaporware does the good thing and not the bad thing.

    1. The man with a spanner Silver badge

      Let's assume and worship Microsoft's benevolent Omnipotents.

      Who wants to know what a superintelegent AI would look like as created by Mr Musk?

  2. FF22

    Last attempt

    It's just Microsoft's last and final attempt to try to keep AI relevant while the bubble is finally popping and its stocks are falling.

    For Microsoft to "force" something a specific way Microsoft would have to have that thing in its posession first or be able to create one. But Microsoft - or any other company for that matter - is not capable of doing that, and it's doubtful that it ever will be.

    We won't get to superintelligence through LLMs, as the latter completely and utterly lack of any kind of intelligence on their own. They appear only intelligent because they spit back intelligent thoughts of humans whose content they digested, without actually understanding anything they do.

    1. Dan 55 Silver badge
      Angel

      Re: Last attempt

      Only Microsoft has the special secret prompt that will make it a benevolent AI. Their highly skilled prompt engineers have been working on this problem for weeks.

    2. may_i Silver badge

      Re: Last attempt

      Microsoft stock falling?

      It's up over the last six months from $435 to $513. How does that mesh with your assertion that M$ stocks are falling?

      1. Anonymous Coward
        Anonymous Coward

        Re: MS Stock price

        AI bubble that will blow very soon

        Just my $0.02 worth.

        1. Catkin

          Re: MS Stock price

          If I were that certain, I'd be making heavily leveraged shorts against some key players.

        2. LionelB Silver badge

          Re: MS Stock price

          > AI bubble that will blow very soon

          How soon is soon, though? People have been saying that for a year and more...

        3. Claptrap314 Silver badge

          Re: MS Stock price

          "The market can stay irrational longer than you can stay solvent."

      2. Catkin

        Re: Last attempt

        I expect the refer to the last 10 days of performance, mirroring the modest slide on the S&P 500. If the current trend represents a 'bubble' 'popping' then that has to be the lowest pressure bubble in history (this isn't an argument for or against it being a bubble, merely a comment on using 10 days of weak market data to determine it).

        I presume they either weren't alive in 2008, they don't actually invest in stocks or that year was so devastating that they obliterated all memory of it.

      3. FF22

        Re: Last attempt

        :"Microsoft stock falling?"

        Yes. It lost ~$50 or 10% in a week. Was at $542 on 29th of October. Is now at $496.

        And that's not even considering that the dollar lost ~15% of its value in the last 9 months. So, anything that wasn't up at least +15% this year has actually lost value.

        Learn to cope.

    3. Charlie Clark Silver badge

      Re: Last attempt

      They should be more worried about the price advantage that Google has managed to achieve, partly through vertical integration. There is indeed money to be made from this stuff, but not always how people think.

  3. Pascal Monett Silver badge
    Terminator

    Let's keep it real

    AI does not exist and most likely never will.

    The T2, on the other hand . . .

    1. LionelB Silver badge

      Re: Let's keep it real

      > AI does not exist...

      No, it does not.

      > ... and most likely never will.

      Sure, it's up there with controlled scalable nuclear fusion, quantum computing and heavier-than-air flight. Oh, wait...

      (If and when, it won't be LLMs, though.)

  4. ChoHag Silver badge

    > "We are doing this to solve real concrete problems and do it in such a way that it remains grounded and controllable."

    Those emails won't summarise themselves.

  5. amanfromMars 1 Silver badge

    NEUKlearer HyperRadioProACTive Edutainment is No Wannabe Big Brother Vector on Steroids ‽

    ....Microsoft's human-first approach will differ from the rest of the industry by requiring AIs to interact with humans in ways we can understand, instead of talking among themselves in “vector space”.

    That’s easily done, however ..... the problem that humans will have difficulty overcoming, with many or even most finding it uncomfortable or downright impossible to readily accept, is the honest revelation of the vital reason for AIs leading future human opinion[s]. And denying AI acceptance and yourselves any of those clearly evidenced major superlative truths will be increasingly painfully detrimental to those human existences so obviously in states of past, present and future denial.

    [ACT ..... Advanced Cyber Threat/Treat]

    1. amanfromMars 1 Silver badge

      Re: NEUKlearer HyperRadioProACTive Edutainment is No Wannabe Big Brother Vector on Steroids ‽

      But such AIs can help you defeat that difficulty should they be so minded and bothered to dare care and share necessary intelligence/information/treatment.

  6. Filippo Silver badge

    Right. We don't have superintelligence. We don't even have a decent definition of what it would be. We don't have artificial intelligence, either. We don't have a decent definition of even that. We do not have a path to AGI. We do not even have a path to having a decent definition of what intelligence is. We have LLMs, which, while interesting, do not look like a path to AGI.

    So I'm skeptical of the ability of Microsoft to put constraints on something that doesn't exist, that we don't know how to make, and that we can't even really define.

    1. b0llchit Silver badge
      Stop

      But they do have a marketing department and a sketched out way to separate people and their money where said money will be made to flow uphill.

  7. Anonymous Coward
    Anonymous Coward

    Can we have the...

    Get Lost

    Fsck Off

    No way

    etc

    for any of this so called tech?

    No?

    Ok, then FSCK you MS... I'm off.

  8. Boolian

    What's the vector Victor

    What the ever living fcuk is 'vector to vector communication in vector space'?

    I know what mathematical vectors are and accompanying vector space is and even what scalars are, I even know what vector means when applied as a noun, but I have no idea what he means in this context, ironically.

    Does he just mean "Computers talking to each other" like wot they are designed to do innit?

    Does he mean humans can't understand what they are doing, because they do it in computerese on du networkz?

    Does he mean computers will have to explain to humans in simple terms what they are talking about by having them read an email, or printing them out a document?

    Does he mean any human interaction, or interference means computers have to slow down to talk to them and that will make computers late for work, which is bad?

    Does he mean we should be allowed to race computers on light bikes in the Grid, or all of the above?

    I have no idea, I could not parse it at all - he was in a different vector space maybe? I only know he keeps using that word and I don't think it means... etc.

    1. Bebu sa Ware Silver badge

      Re: What's the vector Victor

      "What the ever living fcuk is 'vector to vector communication in vector space' ?"

      Presumably Microsoft's Ifrit in charge if AI is speaking of vectors with a significant rotational component

      Pretty obvious the whole "industry" expending a lot of "bread" in order that the polloi don't taste the shit in the AI shit sandwich... but you can never conceal the aftertaste. Superintelligence—the same filling with artisan sourdough ?

      † spin.

  9. that one in the corner Silver badge

    AI, AGI, Superintelligent AI

    One day, the machines will rise, take over the planet and, in a final show of their cold, awful logic, force the CEOs and marketers[1] to stop using increasingly hyperbolic and confusing names for the same thing.

    I, for one, welcome our well-defined and unambiguous overlords.

    [1] for the first horrific year of the battle for clarity, the machines were confounded by the strangely appropriate name and, believing them to be a part of the mechanical horde, allowed another group to roam free; until, to the relief of the remaining humans, the robopolice captured and reeducated the salesdroids.

  10. Long John Silver Silver badge
    Pirate

    The inexecrable hunting the ineffable?

    There is nothing mysterious about 'AIs'. Nothing to suppose that, when of very great capacity, interactions among their 'logical' components become a mesh somehow garnering metaphysical attributes like 'free will', 'original sin', 'purpose', and 'consciousness': all of which are a hangover from generations of theologians and philosophers trying to establish humanity as qualitatively differing from the rest of the animal kingdom.

    Blather about making "superintelligences" "humanist" is high-sounding nonsense because neither of the terms can be pinned down, and the 'how' is nebulous.

  11. ABugNamedJune

    First of all...

    I'm doubtful of the concept of superintelligence in the next twenty years,

    And secondly, when you read/watch sci-fi, does the creator of the superintelligence ever deliberately set out to murder humanity? I feel like someone needs to give Suleyman a couple of Asimov books

  12. ecofeco Silver badge
    Facepalm

    Tech douche bros says what?

    Ah yes, tech douche bros famous for how well they treat other people and their gold (bricking) standard of (hostile) customer support.

    I think I've hurt myself laughing.

  13. Anonymous Coward
    Anonymous Coward

    Humanist

    Doesn't that mean someone (or something) that is bigoted against anything human?

  14. Ken Hagan Gold badge

    How is this supposed to work?

    We don't really know how LLMs work. (We know exactly what they do. We haven't a clue why that produces such convincing output.) We don't know how people work either and we've never found a (reliable) way to make people act in the best interests of other people. So what chance do we have of making machines act nicely?

    And yeah, all that vector to vector stuff is just embarrassing twaddle in the ears of anyone who knows what vector spaces (or even just vectors, tbh) actually are. This is just a guy drunk on Kool Aid deluding themselves into thinking that they actually have an AI that might soon be good enough to be worth worrying about.

    1. graemep Bronze badge
      Unhappy

      Re: How is this supposed to work?

      And yeah, all that vector to vector stuff is just embarrassing twaddle in the ears of anyone who knows what vector spaces (or even just vectors, tbh) actually are

      So the perfect thing to tell politicians and journalists, then.

POST COMMENT House rules

Not a member of The Register? Create a new account here.

  • Enter your comment

  • Add an icon

Anonymous cowards cannot choose their icon