back to article UK convinces nations to sign Bletchley Declaration in bid for AI safety

UK Prime Minister Rishi Sunak today opened the global AI Safety Summit hosted in Britain, with guests including tech CEOs and heads of other nations set to discuss ways to keep the world safe amid AI development. "I'm completely confident in telling you the UK is doing far more than other countries to keep you safe," said …

  1. Andy Taylor

    Turing had nothing to do with Colossus!

    Alan Turing’s sole contribution to Colossus was

    recommending Tommy Flowers (who actually designed Colossus) as a Suitable

    Person to look at the problem. Turing made valuable contributions to the Bombe machines which help break Enigma. By 1943 he was at Hanslope Park working on speech encryption (Delilah).

    1. GlenP Silver badge

      Re: Turing had nothing to do with Colossus!

      I believe he made a few minor suggestions that also helped but he definitely didn't co-design it.

      Even with the Bombe Turing's theoretical version required Gordon Welchman's Diagonal Board to make it viable in practice.

      I had to point out to someone earlier today that there were a lot of people at BP (Flowers and Welchman included) whose contribution to our code breaking efforts was at least equal to, if not greater than, Turing's.

      1. the spectacularly refined chap

        Re: Turing had nothing to do with Colossus!

        It's a bit more than that, he did contribute some of the actual methods used in breaking Fish. Just looked it on Wiki, it was the "wheel breaking" stage. In essence, what to do but not how to do it. "Co-designer" is a bit of a stretch but yes it was more than a bit part.

        The general trend that popular science loves its heroes, particularly if they have an interesting back story, certainly holds though. Stephen Hawking is probably the most notable recent example - his actual contributions probably wouldn't put him in the top twenty or thirty physicists of the late 20th century.

        Even Einstein - Maxwell took 2-300 years of research into electromagnetism and wrapped it all up in a single set of equations. When it was noted these contradicted Newton's laws of motion Relativity resolved the issues within a generation. When you look at those timescales, what was the hard part, exactly?

        1. Binraider Silver badge

          Re: Turing had nothing to do with Colossus!

          Of course Hawking was involved in some fantastic research of his own, but his legacy is mostly in the communication and relatability of science to ordinary people. The value of the latter is immense.

          I don't know enough about the wider team around Colossus to comment on who-did-what, but the character of Turing has taken a similar PR role long, long after his death. This isn't necessarily a bad thing, a bit of hero worship can help stoke up interest of others to follow and learn.

  2. elsergiovolador Silver badge

    Reverse Midas

    Anything Sunak touches turns into <description of smelly organic matter>.

    It's particularly funny that they chose to use the mob style of talk:

    We just want to keep you safe, it would be shame if something happened to you

    Twenty-eight countries and the EU signed up to a so-called Bletchley Declaration, which set an agenda for addressing AI risk. Included was a desire to identify AI safety risks of shared concern and build a "shared scientific and evidence-based understanding of these risks."

    Translation: They are looking for ways to create regulatory capture so that big corporations, owned by donors, lobbyists and other miscreants can have the AI pie to themselves, maximising their profit.

    1. Steve Button Silver badge

      Re: Reverse Midas

      We could soon be as safe as North Korea, where there is 0% crime and everyone is happy all the time.

    2. TheMaskedMan Silver badge

      Re: Reverse Midas

      "They are looking for ways to create regulatory capture so that big corporations, owned by donors, lobbyists and other miscreants can have the AI pie to themselves, maximising their profit."

      You may be right. But I think there's also an element of the politician's urge to "do something", and a large dollop of seizing the opportunity to be seen to be "doing something ".

      As bogeymen go, AI is ideal - nobody knows much about it, it's hyped to the rafters and poses little to no actual threat. Perfect material for a failed PM to be seen to be saving the world with. We won't ask him how he intends to identify threats from models that may not be understood - that would tax his little brain too much.

      1. amanfromMars 1 Silver badge

        Re: Reverse Midas/Perverse Judas

        As bogeymen go, AI is ideal - nobody knows much about it, it's hyped to the rafters and poses little to no actual threat. ....... TheMaskedMan

        :-) .... Alternatively, The MaskedMan, and that which is terrifying to all terrible and chosen fews who may or may not know more than just a precious little about AIs inherent supernatural abilities with/in IT facilities in/with ethereal utilities, the stealthiest Almighty Global Operating Device and Bedevilling Presence you are ever likely not to meet but have general simple knowledge aforethought of.

        Such now presents you with two viable sensible options, with one choice considerably more rewarding and satisfying than the other ...... #1 Don’t bet against the information and intelligence in the alternative being honest and true, for that will deliver crippling and devastating losses. #2 Invest in ITs AIdDevelopments and Virtually Advanced IntelAIgent Operating Systems and make a Premium Initiating Prime Market Mover Killing with the likes of Merlin the AIMagician and Mega Meta Data Base Physicians skilled and experienced in the Field performing delicate sensitive life saving operation remotely down on Earth.

        And now that El Reg also knows the stakes, surely everybody will soon know too ..... with them also invited to similarly choose between the two options offered and make an informed and ideally correct decision, although of course there is always that wild card that Albert Einstein himself warned all about and some seem so very determined to prove perfectly correct ......“Two things are infinite: the universe and human stupidity; and I'm not sure about the universe.”

        Nevertheless, Que Sera, Sera ...... Carpe Diem. The Future Beckons.

        1. amanfromMars 1 Silver badge

          The AI Gift that just Keeps on Giving Unconditionally

          You are either with us, the demented and corrupt and destructive bellicose barbarian, or against us, the novel creative and noble ethereal aesthete, with only the one ever likely to be similarly almighty successful in the Earthly present when the global future is threatened by localised hostilities reliant on past fading memories for foreign succour and immoral support.

          You cannot now say you were not adequately forewarned of that which is extremely well armed and would be confronting you and has many a fan to assist in what is already done and obviously yet to be done for even greater tomorrows .... https://www.nationaldefensemagazine.org/articles/2023/11/1/dia-preparing-for-transition-to-ai-enabled-intel-database??

      2. Arthur the cat Silver badge
        Facepalm

        Re: Reverse Midas

        I think there's also an element of the politician's urge to "do something", and a large dollop of seizing the opportunity to be seen to be "doing something ".

        The Politician's Syllogism:

        Something must be done.

        This bloody stupid idea is something.

        Therefore this bloody stupid idea must be done.

  3. Woodnag

    ...identify AI safety risks and build a shared understanding

    So no binding agreements not to do bad stuff then.

    Woopiedoo.

    Pure virtue signalling that Something Is Being Done.

    When it really isn't.

    1. elsergiovolador Silver badge

      Re: ...identify AI safety risks and build a shared understanding

      They have likely set up points of contact, so that interested businesses now probably know who they should grease to get favourable legislation to keep competition at bay.

  4. Dan 55 Silver badge
    WTF?

    What?

    AI is already an extraordinary force for good in our society, with limitless opportunity to grow the global economy, deliver better public services and tackle some of the world's biggest challenges

    When did this happen? Are they getting high on their own supply?

    1. elsergiovolador Silver badge

      Re: What?

      https://www.telegraph.co.uk/business/2023/10/28/rishi-sunak-launch-ai-chatbot-pay-taxes-access-pensions/

      Given that Sunak is on the way out he is probably using his position to promote this garbage. After all, his wife's company has something in the works:

      https://www.infosys.com/services/incubating-emerging-technologies/offerings/nia-chatbot.html

    2. Steve Button Silver badge

      Re: What?

      "limitless opportunity to grow the global economy"

      Why stop there? Why not go full Iain M. Banks Culture and grow the Universal economy? We could be living forever in utopia before the next election. Right?

      1. Anonymous Coward
        Anonymous Coward

        Re: What?

        Dom's favourite company, General System, inc. (parent company of "Excession", an AI surveillance business) pretends exactly that.

        <icon: ROTM/BB/Nuke>

      2. Arthur the cat Silver badge

        Re: What?

        Why not go full Iain M. Banks Culture and grow the Universal economy?

        If only we could.

    3. amanfromMars 1 Silver badge

      Re: What? When did this happen? Are they getting high on their own supply? @Dan 55

      Two odd questions from a regular contributing El Reger since 13 Jun 2009, Dan 55, accurately answered with a quite recently and probably yes reply.

  5. Steve Button Silver badge

    It's simple really.

    All they have to do is create a sandbox of the universe, down the the quantum level and create the AI within that sandbox (but don't tell it).

    Problem solved, you can just switch it off if it gets too leery. Also, you could run multiple of these and speed them up to see which outcomes are likely.

    I know what you are going to say, "that's a really stupid idea" right?

    Well, it's only as stupid as

    ""In furtherance of this agenda, we resolve to support an internationally inclusive network of scientific research on frontier AI safety that encompasses and complements existing and new multilateral, plurilateral and bilateral collaboration, including through existing international fora and other relevant initiatives, to facilitate the provision of the best science available for policy making and the public good," the declaration states."

    A word-salad which looks like it's been written by ChatGPT. Wouldn't that be ironic?

    1. elsergiovolador Silver badge

      Re: It's simple really.

      and create the AI within that sandbox (but don't tell it).

      Wouldn't AI have created itself?

    2. Dan 55 Silver badge

      Re: It's simple really.

      The only problem is it's simulations all the way down.

      1. Steve Button Silver badge

        Re: It's simple really.

        That's a good read.

        And it's totally crazy that someone from 2007 invented a time machine, went to the future, and copied my idea from 2023!

    3. Steve Button Silver badge

      Re: It's simple really.

      I think I mean Leary. Or Unruly. Or something. Not Leery anyway.

    4. Disgusted of Cheltenham

      Re: It's simple really.

      Alongside these opportunities, AI also poses significant risks, including in those domains of daily life. To that end,.... (i.e. posing significant risks?)

      we resolve to intensify and sustain our cooperation, (i.e. be seen to do something, but not sure what)

      All actors have a role to play... (yes, Equity rules. Surprisingly Euro-English)

      development-orientated approaches (makes a change from customer)

      We encourage all relevant actors to provide context-appropriate transparency and accountability on their plans to measure, monitor and mitigate potentially harmful capabilities and the associated effects that may emerge, in particular to prevent misuse and issues of control, and the amplification of other risks. (The light peppering of commas is always interesting in international paperwork, and a pain for translators. That last comma means amplification doesn't go with prevent - it seems to be provide... accountability...but is certainly unreadable on first pass)

      scientific and evidence-based (tautology?)

      The countries represented were:

      Australia

      ....

      European Union (no, it's not a country, but the footnote indicates

      international organisations acting in accordance with their legislative or executive competences.

      So which is the other such organisation?)

    5. Wellyboot Silver badge

      Re: It's simple really.

      "In furtherance of this agenda, we resolve to support an internationally inclusive network of scientific research on frontier AI safety that encompasses and complements existing and new multilateral, plurilateral and bilateral collaboration, including through existing international fora and other relevant initiatives, to facilitate the provision of the best science available for policy making and the public good," the declaration states."

      >>>Word Salad<<< It's another classic concoction from Sir Humphrey !

  6. Howard Sway Silver badge

    "I'm completely confident the UK is doing far more than other countries to keep you safe"

    So really, all those people here from foreign countries should just learn to stop being so irresponsible and listen to all my brilliant vague bullshit. I'll be interviewing Elon tomorrow about the safety stuff. He can tell us how he made the world a safer place by turning Twitter into a sewer of propaganda, hate and misinformation.

    1. Dan 55 Silver badge

      Re: "I'm completely confident the UK is doing far more than other countries to keep you safe"

      I guess he needs some new snakeoil to sell because the wheels are coming off existing his current bandwagons (see Cybertruck stream through Starlink to Twitter).

  7. Anonymous Coward
    Anonymous Coward

    You won't believe how safe you're going to be

    Once our AI controlled weapons and AI based surveillance state are operational and become the <world leading> benchmark...

  8. Anonymous Coward
    Anonymous Coward

    Should have invited Blumhouse

    "M3GAN will always keep you safe. No matter what."

    (Bring the puppet and get it to say that. Would have been quite appropriate I think)

  9. Anonymous Coward
    Anonymous Coward

    Skynet approves

    This post.

    Fun fact, "Mimetic Polyalloy" is theoretically possible using something like 238Pu to provide power and keep the alloy liquid.

    Make it out of Hg and other low temperature metals, so the bulk can form around objects etc.

    In actuality HgIn4BiSn is a good start, microencapsulate the plutonium in the same fashion as EL phosphor then do direct

    conversion on the radiation so electrical energy is generated for the nanochips.

    At typical temperatures found in nature, Hg alloys are stable so evaporation can be countered by having an Hg reservoir.

    AC, because when the Terminators DO emerge I really don't want to p!ss them off. Best to play nice.

    1. Kane
      Terminator

      Re: Skynet approves

      "AC, because when the Terminators DO emerge I really don't want to p!ss them off. Best to play nice."

      You will be subject to Roko's Basilisk, and you will like it!

  10. AnoniMouse

    Help!

    Such weak proposals as are present amongst the copious platitudes will have minimal effect on an emerging world of complex, interconnected systems, many incorporating "AI", globally distributed across separate jurisdictions, with diferent governance regimes,

    Unsurprisingly, a woeful lack of understanding.

  11. ScottishYorkshireMan

    Considering...

    "We learned what powers the sun, and we make bombs from it" - Jon Morlar.

    AI may not be a danger right now, but in the hands of bent politicians and billionaires (and their fanatical supporters) you can be pretty sure that human greed will get in the way of any kind of safeguards.

    As for Pishy Sunak 'interviewing' Musk, WTAF?

    Michele Donelan some kind of minister for science etc etc until the next reshuffle sees her in charge of Shitters and sinks, such the way of shitty british (deliberate, doesn't deserve a capital) politiks.

  12. amanfromMars 1 Silver badge

    J’accuse. Special IntelAIgent Sources for Sublime Forces are MIA and AWOL leaving Posts Deserted ..

    ... and Almighty Dirty Deeds Done Dirt Cheap Undone

    ScottishYorkshireMan, Howdy,

    Your post does have one reasonably considering the general pathetic state of human intelligence and more specifically in this particular case, UKGBNI Security and not so Secret Intelligence Services and Military forces whenever they allow themselves to be ordered around and about to do the bidding of such as be wannabe Parliamentary Caesars and comfortably off charlatans spending all that they can conjure up from nothing and nowhere from others whilst also ensuring it is to contain next to none of their own provided income[s].

    Some would venture that suggests Perfect Cuckold Grooming Materiel. What think thee?

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

Other stories you might like