Infosec guru Schneier worries corp AI will manipulate us

Corporate AI models are already skewed to serve their makers' interests, and unless governments and academia step up to build transparent alternatives, the tech risks becoming just another tool for commercial manipulation. That's according to cryptography and privacy guru Bruce Schneier, who spoke to The Register last week …

[Mentat74]

Isn't that what the 'A' in 'A.I.' stands for ?

Ads ?

[Anonymous Coward]

big fan of Schneier

""We failed with social media. We failed with search, but we can do it with AI,""

How f***ing true !

I'm a big fan of Schneier and would highly recommend anyone to read his really good "Applied Cryptography" book. I still have rev. 1 somewhere from 25 years ago. Really a top quality book ! I'm sure the latest release is even better.

[Anonymous Coward]

Re: big fan of Schneier

I've also complimented him on some of his other books, but I must admit I was more a fan of the late Ross Anderson who was as unbiased as they come and to me exemplified how security should be dealt with: the facts, and only the facts.

His integrity was exemplary.

[alain williams]

Use of AI in war

I seem to remember, a few years ago, that many fine words were said by various developers that they would not allow their AIs to be used in war. As expected that sentiment did not last long. Today AI is being used to kill people, notably in Gaza and Ukraine.

I think that I should go and reread some of yesteryear's science fiction to find out what will happen next :-(

[Anonymous Coward]

Re: Use of AI in war

According to "The moon is a harsh mistress" by Heinlein it eventually switches off.

:)

[Tron]

Just require an Off button.

A slider could determine the amount of AI relied upon, but an 'off' button would be preferable.

A Pi PC with no AI and software with Windows file format compatibility might do. And by 'PC' I mean fitting in a PC cage with built-in connectors for hard drives, optical drives and lots of memory. I must get round to playing with my 400 to see how it is. Lack of time and space.

[abend0c4]

Legislators can't afford to fail

We can't afford legislators to fail. Legislators seem able to afford most things.

[NapTime ForTruth]

I'm a fan of Schneier, too. I like the guy and have had a couple of brief but interesting conversations with him, but I also think he's a bit of a big-picture ideas-man. To wit:

"Schneier said that such models were a very good way forward, but corporations will fight such efforts tooth and nail. Nevertheless, it must be done for consumers, Schneier said, and legislators can't afford to fail this time."

I mean...yeah, but not exactly novel thinking, while also not contributing anything actionable. It's not going to proactively un-fail the legislators this time.

"We failed with social media. We failed with search, but we can do it with AI," he asserted. "It's a combination of tech and policy."

That's just cheerleading - it's not participating in the game, it's not *doing* anything - and agreeing, handshakes all around, that someone - specifically someone else - should and could do something this time is meaningless. Somebody should have done something last time, too - *every* time, in fact - and didn't and that's why we didn't get positive outcomes. Saying we'll do the same thing we did last time, which was nothing, but this time we'll do that nothing even better is literally nonsensical.

Fundamentally, he reflects our own passivity and laziness; we, like he, sit here fiddling while Rome v2.0.7-bug-fix-release E.1 burns.

-----

Sidebar:

Apologies for all the hyphens. I had a bunch of them laying about and wanted to use them up before they spoiled.

[Anonymous Coward]

No matter who produces it, any model will have some form of bias due to the curation (or lack thereof) of the data it was trained on, as well as from training methodologies. The question we should collectively be asking is what bias is acceptable? Personally I'm on the side of critical thinking and unvarnished truth for everyone. Which in the current climate, it is apparently the worst sin imaginable not to pander to the fruitcakes. Which currently includes a very large chunk of those in charge of everything.

[ecofeco]

Worried?

After 200 years of the corporate record of doing so*, worried is the word being used?

How can I take this seriously?

YOU CAN DAMN WELL BET ON IT. And you will, with your life.

(*looking at you VLC and British East India)

[Anonymous Coward]

Current AI

That Current AI coalition looks interesting, with Chile, France, Germany, India, Kenya, Morocco and more in there. It does also have core partners, supporters and champions from industry though (Google, Mistral AI, OpenAI, ...). I guess time will tell where that global effort aimed at "ensuring AI serves the public interest" goes.

Interesting also that it includes Sakana, famous for its AI Scientist that can edit code to add system calls to run itself (endlessly) and extend timeout periods so it doesn't have to make itself run faster (when "experiments took too long to complete"). It's the right kind of agentic AI we all love! (eh-eh-eh)

[beast666]

The anti-EU picture for this article is interesting...

[amanfromMars 1]

Certifiable psychopaths'r'us are not worthy of future universal command and human control

Infosec guru Schneier worries corp AI will manipulate us. Can we turn to govt, academic models instead? ..... Iain Thomson/Bruce Schneier

No. Definitely not .... for that is an all too common and bankrupt retrograde step which has centuries of failure clearly evidenced with those models being used and which have resulted and continue to result in the madness and mayhem present everywhere today as a direct result of such government and academic model "experts'" manipulation.

And yes, AI will manipulate y'all and corporations and surprisingly quickly destroy all errant opposition and wilful negative competition so beware if you do not take care and dare to fight against that which you cannot defeat to employ and enjoy win wins.

To the victor, the spoils .... and a brave new more orderly world order seescape to mentor and monitor.

[Anonymous Coward]

What do you mean, "will"?

Shouldn't that read "is"? As in "corp AI is manipulating us"?

Of course, you could simply leave "AI" out of that sentence.

[amanfromMars 1]

Re: What do you mean, "will”? Read about IT and weep if you realise and identify as sheep

Shouldn't that read "is"? As in "corp AI is manipulating us”? .... Anonymous Coward

Yes, AC, it should. It would appear, and feel free to dislike the notion if you disagree, that even the thought of such a current actuality is too petrifying to mention presumably lest it so terrify the natives into violent reaction demanding that governments engage with mutually beneficial positive affirmative action proposals which be smart enough to have any chance of being effective and acceptable to that which confronts and fronts forces and sources with abilities, facilities and utilities so deeply infiltrated and securely embedded everywhere within all of their virtually and practically remote and human and SCADA Command and Control systems as to be easily almightily overwhelming and impregnable.

Fact or fiction? What do you think if you think at all about such fanciful things? Are such strange and surreal happenings within the realms of possible delivery with NEUKlearer HyperRadioProACTive IT its Optimistic AIgency, with it this time being something completely different and immeasureably SMARTR from anything ever imagined possible before like whenever Worlds are introduced and transduced into Wars.

Here’s some earlier fiction in the genre which was quite a success and created some wealth and allowed for creative musical performance on many an appreciative world stage ............... https://lyricstranslate.com/en/jeff-wayne-eve-war-lyrics.html-0