When a billionaire boy wonder drives the enterprise, all kinds of *bs* is going to be inserted where it doesn’t belong.
We can’t just allow the industry to do what it wants to do, because they will maximize profits. We need “bright line” standards that protect kids, that protect workers
Overseen by an independent government agency.
What government agency even understands AI? None of us understand the full power of disruptive AI. Privacy. Kids. Jobs. Yes, those are the buzzwords.
We’re all pro-guardrails. But the rest of the world is not going to be playing with us. Do you think our government oversight is going to put limits on China, Russia, and dark agents in the Middle East? We can pontificate all we want. We can *all* see how this technology is being used to build deep fakes, video edits completely made up, and steal personal, private, and professional information. It’s odd that AI created an alliance between myself and Rob McElhenney, but it’s an ai hallucination. The LLMs suck in all the information they can scrape off any and every digital source of data, even if the data is supposed to be private an protected.
Ask an AI about Elon Musk and you’re going to get a lot of fake stories. The problem is identifying what is “made up” and what is fact. Also, determining where the data came from will be big business. Forensic AI will look at all data sources and determine where the false information came from. But, what if an army of bad actors is feeding, purposefully feeding the LLMs of Open.ai and Bard false information? What if huge manifestos of lies are being presented to the web as #truth? (Fox News alternative facts mixed in with the actual facts.)
ChatCPT is actually getting dumber. As the training is taking place, by everyday users like myself, we are feeding the *machine* data that it can use in the future. Hmm. Upload that confidential financial report on the skyrocketing cost of batteries for EVs, and you’ve got a case for insider information when the AIs suck down and manipulate your numbers. The scariest part of AI at the moment is the volume of information it is vacuuming up from all available data sources. No one can really track where the information is coming from, and once it is washed (shared by a ChatGPT user or Lang Chain bot that pulls on GPT’s data) the false information is out there.
As Google and Bing add AI to their search algorithms, and Siri gets an AI uplift, how are we going to filter the fake news (ai generated falsehoods) from the facts? Already, there are sniffers being trained on spotting AI-generated content, but the process is more speculative at this moment. Will Open.ai’s “nanny bot” actually report on itself and remove false information? What about illegally obtained information?
The new WikiLeaks is going to be LLMs. Feed them unlimited data from Fox News and you’re going to start altering the “reality” of the LLM. Non-human intelligence is not capable of self-regulation. It’s like the Three Laws of Robotics penned by Asimov in the Foundation Series.
It seems to me, we need a more holistic approach to reigning in the machines before they Lang Chain themselves together into Skynet. That’s the banter around Capital Hill today. “The danger of AI.” Protecting “children and workers” from the dangers of the ai invasion. Protecting business secrets and private information from the ai redistribution of data. The people assembling to teach the senators about AI are the AI businesses themselves.
Can Nvidia (the current king of AI chip power) be honest about the threats of LLMs running on his silicon? Can the CEO of ChatGPT extol the benefits and how they outweigh the risks?
Today on a financial show about “the markets” a host said, “This is just like 10 years ago. Just think if the government had stepped in and regulated social media.”
What? Even this “financial wiz” doesn’t understand the past or the current moment. We’re all looking for the “business case” and profits for corporations and shareholders. Everyone heavily invested and now much wealthier due to Nvidia’s doubling and tripling of value, is invested in seeing AI expand and grow with unlimited storage, compute power, and access to all information it can scrape from the internet.
If the internet is now your personal LLM, you’ve got to build personal tools to help you navigate the faux news from the facts. Just like watching TV, you can’t comprehend how one network’s view of the world is do dark and antagonistic. It’s hard to imagine a more bitter and vile stream of words coming from Fox at this time. The doublespeak happening unfettered in the right-wing media complex is feeding just as much data into the bots as the more reality-based news channels. But how is the LLM and its trainers to decide what is CORRECT, UNKNOWN, UNVERIFIED, or FALSE? And then at what lever up the chain of command does the decision come down to “keep the illegal data lakes” for future opportunities?
The future is here. It hasn’t brought us flying cars or neuro-linked video gaming. But we do have a new iPhone with a USB-C charging port. We have semi-self-driving Teslas. We have billionaires warping our worldview with their own form of double-speak. We have jobs that are killing so many of our souls. And we have the lowest levels of unemployment in decades.
Why aren’t we all happy in our prosperity and techiness? Why won’t the newest and most powerful iPhone yet, make us happy? Maybe we need to spring for the Pro Max? The size of early tablets, I mean, if you’re always on your phone it might as well be the fastest and largest you can afford.
The iPhone is the Ferrari of phones. Having the latest iPhone is a demand that bears little or no relevance to the “new iPhone features.” The reason Apple will sell a boatload of iPhone 15s is because it’s an easy and attainable status symbol for everyone. Your carrier will help you into the latest phone, and your bill will only go up by $25 a month until you pay it off.
The mobile phone upgrade cycle for kids is relentless. The first kid with an iPhone 15 will be a god at school. And those iPhones are showing up earlier and earlier in elementary school. See, we as parents have no boundaries for our kids. A new iPhone is an easy concession. And the carrier gives it to me for free. Well, okay, no interest, but nearly the same thing. If I pay it off in 24 months. It’s a trap. They know we will not make it to the 2-year mark. The iPhone comes in new colors and models just before the holiday season. Wonder about that?
It’s easier than thinking. Give your kid a newer iPhone for Christmas. It won’t even cost very much. You’re still paying on the last phone, and they will simply give you a new one, let you trade in your old one, and if you’re a real winner, your monthly interest payments may even go down. But here’s the thing. This kind of purchase is more like a lease. Kids and many adults will never stop lusting for the latest iPhone or the latest laptop. This is where Apple dominates both markets in terms of revenue, and profitability. A lower market share in computing doesn’t worry Apple at all. Would you rather sell 10 laptops for $650 each? Or 3 laptops at $2,500 each? Apple will continue to chug along. And when the time comes they will fight for Siri to keep all her data (found, scraped, stolen, and provided).
The tech leaders want to see this AI boom pay dividends. The congressmen on these committees are wealthy investors who probably own a lot of Nvidia, Apple, and Amazon stock.
The AI golden egg fast becoming a nightmare if you’re interested in limiting the power and reach of a system we don’t fully understand, much less, understand how to set up limits and guardrails. If we knew what the boundaries were, how would we go about enforcing them? If China wants to inject our ChatGPTs with an unlimited stream of illegally obtained data dumps who’s going to regulate them? Who will firewall off their AI and risk falling behind the companies that are exploding with dark data, ai hallucinations, and malware-trained LLMs.
There’s very little protecting ChatGPT’s public experiment from loading, processing, and regenerating lies. And what’s the consequence? A news story about Hillary Clinton being named the 45th president of the United States. This happened at the unveiling of BARD, Google’s AI LLM. In the live demo of BARD a plump lie (many of us wishing for this to have been the outcome of the 2016 election) of Hillary’s victory was broadcast to the world with the confidence of a robot.
Resetting Our Relationship to Robots
That Roomba gathering dust in your closet is an example of ai promise that has not been fulfilled. Yet! I am sure a Roomba.ai is on the near horizon. It’s still not going to help my Roomba not get stuck under the couch or smear over a doggie mess in the hallway. And what about self-driving cars for the rich vs. the rest of us?
Last week “Elon Mode” was discovered on some Teslas. Regulators are now looking into yet another Musk controversy. In this one, a software “feature” was installed on some Teslas that would allow you to bypass the self-driving safety protocols. In “Elon Mode” the driver could essentially climb in the back seat and make love to a porn star while the car was stuck in mid-afternoon traffic. The safety mechanisms under “autonomous driving” mode require the driver to be proper seat. A scanner is watching the eyes of the human co-pilot to make sure they are always watching the road. Can you imagine the audacity of a man who orders an override to be created? Then imagine this same *boy genius* begins selling this illegal autonomous driving option to his friends. Regulators are now looking to discover how many Teslas purchased the Elon Mode hack. The number is nonzero
If Elon will alter his robot to do bad things, what’s to stop him from shaping his AI (“we’re building the most-powerful AI) to do bad things as well? Like Twitter, when a billionaire boy wonder drives the enterprise, all kinds of *bs* is going to be inserted where it doesn’t belong.
And that’s the moral of the story. Two sides to this battle: Good = truth, public information only, purge LLMs away from false information; Evil = newtruth, lies, all available data regardless of validity or privacy, LLMs are trained toward false information and are corrected back to the lie when corrections are attempted. Get this, if Elon controls our self-driving cars, our security uplink satellites, our second largest social media information (true and false) input, and the rocket ship seats to escape the ravaged Earth in 20 years, you’re looking at a new Terminator. He drives an ugly cybertruck with a smashed rear window, and he dances like Elaine on Seinfeld, but yes, he’s the GENIUS we are following. EVIL GENIUS? Who knows? It’s too soon to tell. MISGUIDED NARC GENIUS? 100% But with that much money even the government is a bit cowed in how it approaches the lies and deceptions going on at Tesla and with the Tesla customers.
This is not going to work out unless we revise the Three Laws of Robotics. Envisioned 70 years ago by Isaac Asimov. We’ve got to reset our boundaries and consider placing all ai technology in a sandbox environment, where the information in and out is monitored and filtered before it is allowed to escape into the real world of Google Search, Word’s Writing Assistant, and the other 10,000 ai offerings heading toward your car, your browser, your kid’s iPhone and Apple Watch.
It’s our use of ai that needs to be regulated. But who will regulate the cash cows and billionaires?
The Problem with ai regulation efforts today is how they are worried about “privacy” as the biggest threat of ai’s looming awakening. Maybe they didn’t pay attention in Terminator. For sure we have no idea what the LLMs are doing behind the scenes, our cybersecurity systems have already been subverted. Our Battlestar Galactica moment is near at hand. We’re all worried about P. K. Dick’s Gattica. But we’ve already passed that tipping point. Today, it’s going to be profits over privacy. Tomorrow it’s going to be “Let’s allow them unlimited access to share information with each other. The limits and guardrails are outdated and unhelpful.”
With the trillions at stake, even Mr. Musk is awed by the rising force. He’s going to work to keep the shareholders happy and the cash in his coffers. He doesn’t really care much about regulators and restrictions. He knows, that with enough money, those rules don’t apply to him or his empire of wealth.
Keep it classy out there in Westworld.