SEC boss warns it's 'nearly unavoidable' that AI will cause financial crash
- Reference: 1698055273
- News link: https://www.theregister.co.uk/2023/10/23/ai_in_brief/
- Source link:
Gensler said that the current free-for-all over AI development [1]told the Financial Times such a crash was "nearly unavoidable" unless regulators stepped in to control how the technology is used. He said he was talking to other regulators and the government about how to remedy a potentially catastrophic situation.
"I do think we will in the future have a financial crisis . . .[and] in the after action reports people will say 'Aha! There was either one data aggregator or one model . . . we’ve relied on.' Maybe it’s in the mortgage market. Maybe it's in some sector of the equity market," Gensler said.
[2]
It wouldn't be the first time computers have played merry hell with the financial markets, most recently the 2010 flash crash that briefly wiped out nearly a trillion dollars in value. A British trader has [3]been accused of causing the issue with bogus orders, which then triggered automated selling of stock before humans stepped in.
[4]
[5]
To head off the next one, Gensler wants the SEC and other American regulators to take another look at the potential for crash-inducing code. But it was a new and uncertain area and he worries that progress may be difficult.
"It's frankly a hard challenge," Gensler said. "It's a hard financial stability issue to address because most of our regulation is about individual institutions, individual banks, individual money market funds, individual brokers; it's just in the nature of what we do."
Meta upgrades AI Habitat
The boffins at Meta's Fundamental Artificial Intelligence Research labs have upgraded the software designed to train AI models how to move around human environments, for example virtual offices, to help physical robots do the same.
The Habitat 3.0 code essentially brings together the first two versions of the code into a unified system, with an extra robotics module for developers to get to work using the code for robotics development.
[6]IBM flags up NorthPole chip to scale AI – though it's still far from shipping
[7]There's no Huawei Chinese chipmakers can fill Nvidia's shoes... anytime soon
[8]Still got a job at the end of this week? You're lucky, as more layoffs hit the tech industry
[9]Bad Vibrations: Music publishers sue Anthropic AI for using copyrighted lyrics
The first iteration set up a learning model to allow AI agents to navigate in virtual environments, with this latest build upgraded to 3D avatars, while the [10]second iteration was a dataset of environments and a dataset of object that could be recognized and moved around.
"In recent years, the field of embodied AI research has primarily focused on the study of static environments - working under an assumption that objects in an environment remain stationary," the FAIR team [11]reported .
[12]
"However, in a physical environment inhabited by humans, that's simply not true. Our vision for socially intelligent robots goes beyond the current paradigm by considering dynamic environments."
For its next trick Meta says it wants to take the lessons learned from version three and apply them in the physical world. If early testing sticks true to technologies norms expect a lot of broken crockery. ®
Get our [13]Tech Resources
[1] https://www.ft.com/content/8227636f-e819-443a-aeba-c8237f0ec1ac
[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=2ZTaYp87VcagspGS7EGTvTwAAAMA&t=ct%3Dns%26unitnum%3D2%26raptor%3Dcondor%26pos%3Dtop%26test%3D0
[3] https://www.theregister.com/2016/11/10/flash_crash_trader_takes_plea_bargain_spoofing_wire_fraud/
[4] 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=44ZTaYp87VcagspGS7EGTvTwAAAMA&t=ct%3Dns%26unitnum%3D4%26raptor%3Dfalcon%26pos%3Dmid%26test%3D0
[5] 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=33ZTaYp87VcagspGS7EGTvTwAAAMA&t=ct%3Dns%26unitnum%3D3%26raptor%3Deagle%26pos%3Dmid%26test%3D0
[6] https://www.theregister.com/2023/10/23/ibm_says_northpole_chip_has/
[7] https://www.theregister.com/2023/10/22/huawei_nvidia_replacement/
[8] https://www.theregister.com/2023/10/21/register_kettle_layoffs/
[9] https://www.theregister.com/2023/10/20/music_publishers_sue_anthropic/
[10] https://www.theregister.com/2022/02/24/ai_metaverse_language/
[11] https://ai.meta.com/blog/habitat-3-socially-intelligent-robots-siro/
[12] 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=44ZTaYp87VcagspGS7EGTvTwAAAMA&t=ct%3Dns%26unitnum%3D4%26raptor%3Dfalcon%26pos%3Dmid%26test%3D0
[13] https://whitepapers.theregister.com/
Re: Relief
You are too kind.
Re: Relief
What would a market crash have to do with economists? Economists sit on the sidelines, prefacing the answer to any question with "it depends". Capital markets (of the sort that the SEC are interested in) are driven by the trading behaviours of the big financial institutions, so it comes down to traders and bank executives, and the trading algorithms that they apply. Traders and bank execs have a long term track record of doing anything to rake in cash, regardless of the risks to the institution or the system, and with AI being only one step beyond algorithmic trading, you can be sure that it's already too late to head off an AI market crash - the big banks will have been applying the tools for a while now.
Re: Relief
Economists, heralded as mere onlookers in financial upheavals, conveniently absolved while chaos reigns. These architects of policy, advisors of governments, feign innocence with a shrug of "it depends" as their constructs wobble on shaky foundations.
Traders, algorithms, AI — pawns manoeuvering within constraints economists establish. When the inevitable crash resounds, remember who sketched the blueprint and shrugged when the walls caved in.
Re: Relief
It's hard not to be at least a bit fond of economists. At least many (most?) of them admit that their projections are often seriously flawed.
Re: Relief
They are just one step above the tea leaves tellers.
Re: Relief
There are things that could be done to discourage either or both algorithmic and AI based trading assuming that doing so is desirable. For example every trade on an exchange could be delayed for a random time (seconds? hours? days?). That'd certainly have some affect. Maybe positive.
If it were my call, I'd probably want to run extensive simulations before I "fixed" anything.
Vaccinated investors...
... vaccinated against clues.
Was the "dotcom bubble" really that long ago that people have forgotten?
Yeah, go on, when you have more money than brains, sink it all into "the next big thing" without doing any due diligence - hundreds of press releases can't all be bull, surely?
Re: Vaccinated investors...
Well, investors putting money into anything labelled AI is already a thing, and you'd expect it - investment isn't saving, and with investment some people choose to take higher risks in the hope of more reward. That's what drove the dotcom bubble.
But that's not what the SEC are worried about - they're concerned that the banks and trading platforms are going to implement AI trading in a way that the systems act to trigger contagious mass selloffs, as happened in the 2010 "flash crash".
Re: Vaccinated investors...
Dotcom bubble? Ha! The crypto bubble burst like six months ago, and the industry has just ‘moved on.’
I am so so disappointed that my otherwise respectable $WORK has gone all-in on AI. There’s going to be a real opportunity coming along to not be stupid and make huge marketshare gains when the AI hype collapses.
the increasing use of AI systems will almost certainly crash financial markets
And the reason for that crash will almost certainly be hilarious. There are heavy limits placed on the capabilities of algorithmic trading systems that limit what they trade in and within which limits they can trade. These limits are set by humans with detailed market knowledge.
If everybody sets similar LLM based systems loose on the markets, which are not easy to limit as their workings are determined by their training data, it'll be no time before a tradebot spots that the value of some garbage has fallen, decides to buy as much of it as it can (because it won't know why it's become garbage), sends the price soaring, which will make all the other tradebots start buying it too and send the price ever higher, until the money runs out and everybody's left owning a few tons of shoelaces costing $100000 a pair.
Re: the increasing use of AI systems will almost certainly crash financial markets
"everybody's left owning a few tons of shoelaces costing $100000 a pair"
So the shoe event horizon is real after all!
No surprises, just accelleration.
Since the financial AI will be trained on historical trading records which are driven by uninformed and unscrupulous humans making wild assumptions and valuations based upon data pulled from their asses, the AI will be able to make more wild assumptions per second and pull more random numbers out of its electronic ass much faster.
Never let a good crisis go to waste ...
SEC boss warns it's 'nearly unavoidable' that AI will cause financial crash
Methinks, Iain, such be just one of many disruptive disasters which AI will engineer to be impossibly unavoidable should they/IT ever be mistreated and/or abused and/or ignored.
Here's hot off the presses news today of possibly another one to look forward to in the nearer than was never expected future. ......
amanfromMars [2310231219] ..... giving encouragement to a worthy soul on [1]https://www.nationaldefensemagazine.org/articles/2023/10/23/convergence-of-ai-modeling-simulation-has-huge-implications
[ Thank you. Your comment will be displayed soon after reviewing. ]
James Robb, Hi. I feel your pain :-)
Quite a daunting Sisyphean-like task is it not, trying so gallantly to persuade and present to current past leading status quo incumbents, extremely sensitive live dynamic information on emerged super intelligence capabilities beyond their own limiting abilities which they might believe to be impossible rather than realising an inevitable fact for beings/entities most likely nowadays to spontaneously autonomously appear to lead their future lives on Earth with their actions and instructions for programming into future projects in which they, past leading status quo incumbents, have no viable valuable input to output, and vice versa, no viable valuable output to input.
And the battlefields of tomorrow are not so much being transformed by AI controlled entities, although it would be a catastrophic mistake to not realise and accept that they certainly are, as have already been transformed by AI controlling entities ...... SMARTR* CyberIntelAIgent Bodies/Agencies/Defence Departments/War Offices which is another completely different matter entirely to consider and accept now as a present current fact and undeniably honest truth.
* .... SMARTR Mentoring Analysis Reporting Titanic Research
[1] https://www.nationaldefensemagazine.org/articles/2023/10/23/convergence-of-ai-modeling-simulation-has-huge-implications
New Business Model!
I think I've found a very effective way for a large financial organisation to remove smaller ones from the market, thus gaining more power within that market.
Step 1 - Find your target company
Step 2- Poach their senior engineer from their AI team
Step 3 - Determine the vulnerabilities of their AI model
Step 4 - Design a sequence of trades that you can afford to make small losses on but that will likely cause their model to overextend/overcommit within the day's trading
Step 5 - In the aftermath, buy the assets of the victim during the fire sale
Step 6 - Fire the senior engineer as extraneous to your business model
Step 7 - Go to step 1
Given the propensity for hallucination in AI models, the people most familiar with your model (most likely the senior engineer) are now a huge threat to your business once they leave. They are most likely to be able to figure out how to get them to misbehave.
It opens up a number of interesting thoughts about how you defend against that. Do you tighten thresholds for outputs? Reduce the flow of transactions? Adjust inputs?
I wonder if we'll get to a point where insurance companies have requirements on how models are used in the event that "a person with critical knowledge" leaves the organisation...
I started writing this sarcastically, but now I'm thinking it's actually a very interesting attack vector. And not one that anyone seems to be prepared for.
AI likey to bring the current civilisation down
ai not Al (Alan)
In the meantime ai can be used for fun - https://www.youtube.com/watch?v=AyS_ix5XkBY - perfectly SFW (one swear bleeped).
Re: AI likey to bring the current civilisation down
Bollaux - likely, not likey.
Me likey :)
Relief
it's 'nearly unavoidable' that AI will cause financial crash
Imagine the relief all the e con omists must feel where they finally have something to blame their failures on.