AI could crash democracy and cause wars, warns Japan's NTT
- Reference: 1712643307
- News link: https://www.theregister.co.uk/2024/04/09/ntt_ai_apocalypse/
- Source link:
"If generative AI is allowed to go unchecked, trust in society as a whole may be damaged as people grow distrustful of one another and incentives are lost for guaranteeing authenticity and trustworthiness," asserted the Japanese telco in a [1]joint proposal created with media outlet Yomiuri Shimbun.
"There is a concern that, in the worst case scenario, democracy and social order could collapse, resulting in wars," the two entities alleged.
[2]
NTT and Yomiuri Shimbun began a joint study on the ideal governance of generative AI in the fall of 2023.
[3]
[4]
The companies took offense with the lack of accuracy inherent in the technology. They warned that it presents increased danger, as most people with internet access can also access AI.
[5]Mamas, don't let your babies grow up to be coders, Jensen Huang warns
[6]'How do I reset my router' isn't in LLM corpuses. An alliance of telcos wants to change that
[7]Two sats, one customer: Japan's NTT signs up for Amazon's space internet
[8]Japan may join UK/US/Australia defense-oriented AI and quantum alliance
Their proposal concedes that the metaphorical cat is already out of the bag – removing generative AI at this point in time would result in a productivity decline – therefore "the out-of-control relationship between AI and the attention economy" must be confronted.
They assert that as generative AI enters the innovation phase, regulations will become necessary to secure foundational pillars of society – like elections and security.
Suggested regulations include compulsory education for students on the advantages and disadvantages of AI, updates to intellectual property (including copyright) protection, and labelling original content.
[9]
NTT and Yomiuri Shimbun acknowledge that, unlike the EU, Japan does not have a strategic and systematic AI-related data policy. As building one will take some time, "soft laws" that "offer a co-regulatory approach that identifies stakeholders" should be adopted in the meantime.
The proposal also suggests multiple AIs should be available – to keep each other in check and, so that users can cross-reference results and avoid dependence on one particular generative AI product.
NTT's overall take of generative AI is a rare expression of doubt on the technology from the tech industry itself. Most big players are busy tweaking their offerings to capitalize on AI, its equipment, and infrastructure.
[10]
AI-embracing attitudes are present in the likes of cloud providers and chipmakers – but not exactly absent in telcos.
Fellow telco giant Singtel recently kissed the ring of [11]AI-proliferation advocate Jensen Huang, when it [12]announced a collaboration with Nvidia to deliver purpose-built AI infrastructure across Southeast Asia. ®
Get our [13]Tech Resources
[1] https://group.ntt/en/newsrelease/2024/04/08/240408a.html
[2] https://pubads.g.doubleclick.net/gampad/jump?co=1&iu=/6978/reg_software/aiml&sz=300x50%7C300x100%7C300x250%7C300x251%7C300x252%7C300x600%7C300x601&tile=2&c=2ZhURvIkOFE-d7Tbaotm05QAAAIk&t=ct%3Dns%26unitnum%3D2%26raptor%3Dcondor%26pos%3Dtop%26test%3D0
[3] https://pubads.g.doubleclick.net/gampad/jump?co=1&iu=/6978/reg_software/aiml&sz=300x50%7C300x100%7C300x250%7C300x251%7C300x252%7C300x600%7C300x601&tile=4&c=44ZhURvIkOFE-d7Tbaotm05QAAAIk&t=ct%3Dns%26unitnum%3D4%26raptor%3Dfalcon%26pos%3Dmid%26test%3D0
[4] https://pubads.g.doubleclick.net/gampad/jump?co=1&iu=/6978/reg_software/aiml&sz=300x50%7C300x100%7C300x250%7C300x251%7C300x252%7C300x600%7C300x601&tile=3&c=33ZhURvIkOFE-d7Tbaotm05QAAAIk&t=ct%3Dns%26unitnum%3D3%26raptor%3Deagle%26pos%3Dmid%26test%3D0
[5] https://www.theregister.com/2024/02/27/jensen_huang_coders/
[6] https://www.theregister.com/2024/02/27/global_telco_ai_alliance_llms/
[7] https://www.theregister.com/2023/11/29/ntt_amazon_kuiper/
[8] https://www.theregister.com/2024/04/09/japan_aukus_pillar_2_tech/
[9] https://pubads.g.doubleclick.net/gampad/jump?co=1&iu=/6978/reg_software/aiml&sz=300x50%7C300x100%7C300x250%7C300x251%7C300x252%7C300x600%7C300x601&tile=4&c=44ZhURvIkOFE-d7Tbaotm05QAAAIk&t=ct%3Dns%26unitnum%3D4%26raptor%3Dfalcon%26pos%3Dmid%26test%3D0
[10] https://pubads.g.doubleclick.net/gampad/jump?co=1&iu=/6978/reg_software/aiml&sz=300x50%7C300x100%7C300x250%7C300x251%7C300x252%7C300x600%7C300x601&tile=3&c=33ZhURvIkOFE-d7Tbaotm05QAAAIk&t=ct%3Dns%26unitnum%3D3%26raptor%3Deagle%26pos%3Dmid%26test%3D0
[11] https://www.theregister.com/2024/02/27/jensen_huang_coders/
[12] https://www.theregister.com/2024/02/01/singtel_inks_nvidia_deal_to/
[13] https://whitepapers.theregister.com/
No I am!
My name is also Spartacus
NTT's overall take of generative AI is a rare expression of doubt on the technology from the tech industry itself.
Perhaps one of the few voices of sanity within the mad rush to commercialise the latest trend?
No, this is another voice in the choir. Honestly, I wonder if Microsoft paid them to say this.
Have they ever used an LLM? Have you? The things are dumb as bricks. They're easy to spot, especially because you simply cannot get them to shut up, they will turn a thought that can be expressed in two words into a 10 page essay, all while rambling about absolutely nothing.
Actual critics don't inflate LLMs at all, helpful or hurtful, because they see them for what they are; fancy word-suggestion algorithms. Whether it's snake oil or doomsday fanaticism, all of it is inflating the capabilities of a very simple algorithm that genuinely just does not perform well in day-to-day tasks when it comes to talking with a real person.
I believe one of the victims of AI will be Facebook, as there will be millions of fake profiles created continuously which will be difficult to detect as AI and which seem to interact with other users in a manner indistinguishable from a real human. This could undermine trust in the platform and the eventually its scuttling.
Another downside will be that prosecutions of criminals and corrupt politicians will become all but impossible since audio and video footage of their crimes could be argued to have been generated by AI. Over time it will become difficult to distinguish generated from authentic content leaving the courts no option but to disallow the evidence.
Re: Facebook
"difficult to distinguish generated from authentic content leaving the courts no option but to disallow the evidence."
(Not an expert caveat) - I believe there are ways to digitally sign a recording at the time of creation to prove authenticity. Not sure how easy they would be to (a) use and (b) subvert
Quis custodiet ipsos custodes?
"multiple AIs should be available – to keep each other in check"
If it's in their perceived interest (and we have no real idea of how an AI could perceive it's own 'interest'), AIs meant to keep each other in check can still sneakily collaborate... so you need another AI to keep them in check, and then another and it's AIs all the way down.
"Quis custodiet ipsos custodes?" is difficult enough to design in human-based systems even when we know a lot of what drives human psychology, and with hunderds or thousands of years of experience in what doesn't work. What chance is there of designing a system of AIs that really can keep tabs on each other?
Nation Shall Speak Peace Unto Nation
AI could crash democracy and cause wars, warns Japan's NTT
Calls for ecosystem in which AIs keep other AIs in check, and lots more regulation ..... Laura Dobberstein [El Reg]
Of the possibility of Japan's NTT warning proving true, there is certainly no doubt, Laura, with the only real effective preventive solution indeed being greater AIs keeping lesser AIs in check although with a great deal more in the way of revelations rather than regulations as the transparent means of remote command and virtual control.
And most definitely, without any shadow of doubt, if not to be treated as suggested and supplied, a live current presence and existential threat extremely gravely to be regarded because of the discombobulating fact, oft trailed and celebrated in fictions, that there be no available viable human defence/attack vector.
Like this post if you believe I'm an LLM sent to destroy democracy.