Don't worry, folks. Big Tech pinky swears it'll build safe, trustworthy generative AI

Eight big names in tech, including Nvidia, Palantir, and Adobe, have agreed to red team their AI applications before they're released and prioritize research that will make their systems more more trustworthy, the White House tells us.  Today, the Biden-Harris administration boasted it had secured voluntary commitments [PDF] …

[Omnipresent]

Do No Evil

Utter, and complete nonsense. As long as evil exists it will use the tools available, and the rest will have to do their best to stay ahead of the evil. How, exactly, do you NOT keep up with evil?

It has nothing to do with what's right and wrong. It's ALL wrong. It has to do with the root of all evil. MONEY. Money is power, and the monkeys want power. The monkey wants to be king of the mole hill.

[Anonymous Coward]

@Omnipresent - Re: Do No Evil

So, Lone Starr, now you see that evil will always triumph, because good is dumb. -- Dark Hemet.

[An_Old_Dog]

Empty Promises

Some CEO singing the finger-promise song ("Yubikiri genman, uso tsuitara hari senbon nomasu.") is pointless unless there is an organization both willing and able to enforce that promise (e.g., whomever lies [breaks their finger-promise] will be made to swallow one thousand needles and cut off a finger).

[Filippo]

>Generative AI is prone to producing false information that could be manipulated to spread misinformation.

And they propose to fix that how, exactly? At the moment, we don't have the ability to prevent LLMs from producing false information, and we don't even have a theorical model of how this could possibly be done.

>The US government wants Big Tech to develop watermarking techniques that can identify AI-generated content.

I can run LLMs offline on my own PC, and it's all open source. What's going to stop a malicious actor from just disabling whatever watermarks or other mitigation Big Tech put in place? The only way this could work is if the mitigations are baked in the model weights, but (see above) we don't know how to reliably do that and we're not even sure it's possible.

Even that would be a temporary measure, given that the number of actors with the capability to train a LLM from scratch is already almost too large to be effectively regulated, and it will only increase rapidly.

Face it, this genie is not getting back in the bottle.

[amanfromMars 1]

Managing Risk? Oh please, ... whenever the Almighty Stakes are Awesome and Alien? What are They On?

Adobe and Stability AI build text-to-image products, while Cohere, IBM, Salesforce, and Nvidia offer large language models to enterprises. Meanwhile, Palantir and Scale AI are both US government contractors developing and integrating models for the military, which just about covers the gamut of hypothetical Black Mirror episodes. .... KQ/El Reg

:-) Very nearly, Katyanna, but fortunately/unfortunately not quite, and in those sectors, for there are bound to be an encouraging greater number than just one of them, is where all of the leading future specific action is taking place in spaces remotely impacting upon all species and invested entities grounded down on Mother Earth, and exploring in further reaches of the great beyond.

And something else which is certainly really no small thing, and which no one/nothing has yet ventured or appeared to mention loudly and clearly and advise/warn inhabitants and reprobates alike of, is a very attractively addictive and unbelievably rewarding, practically exclusive and virtually real delivery of awesome power and unprecedented energy naturally accompanies and drivers expertise in and experience with AI and ITs Large Language Modelling Machinery ....... and that quite autonomously and quite magically renders to just a very great few, a full and absolute command and control of the vast many via an internetworking chain of remote disguising proxy agents/reactive trojans which you may be assured is a facility and capability not suitable for joint executive sharing with human administrative bodies/capital institutions/socialist ventures with barbaric tendencies all too apparent in recent past evidence, and thus are the vast many fated to be future attentive ignorant spectators rather than destined to be Greater IntelAIgent Game players and novel field leading stars.

That is just the way it is written and to be.

[amanfromMars 1]

Re: Managing Frisky Per Ardua ad Astra Meta Data Base Requirements for the Beyond

And all of the above has really important implications for more than just basic and advanced para-military sectors and conventional military industrial complexes ..... which does appear to be something recognised as critically vital and crucially missing in U.K. Ministry of Defence forces and Secret Intelligence Service sources by at least one Chief of the Defence Staff, Air Chief Marshall Sir Richard Knighton .....

This will have “really important implications” for industry, he said.

“We don't know quite what that requirement looks like, so our traditional model of setting requirements and putting it out to contracted competition won't work,” he said. “In the future, I think we're going to need to see ourselves in government working more closely with industry, focused on delivering outcomes, not necessarily delivering requirements.” ....... https://www.nationaldefensemagazine.org/articles/2023/9/13/uk-military-to-focus-on-outcomes-over-requirements

And either some or many future capabilities/utilities/facilities may have to be accepted by national and international defence forces as being only able to be delivered by proxy services contracted to supply requisite outcomes for a mutually agreeable dynamic price and not egregiously disagreeable extortionate cost.

You can be sure such is what they are guaranteed to be facing supplied by A.N.Others, and either in friendly enough competition or downright hostile opposition, so there’s no escaping future engagement to ensure a dominant present position at the forefront of both novel emergent highly creative and deeply subversive postmodern developments.

cc ..... Chiefs of the Defence Staff, c/o The Defence Delivery Group (DDG), U.K. Ministry of Defence, Whitehall, London, SW1A 2HB, United Kingdom