In a recently published article Bloomberg worried out loud about AI disasters to come. They were referring to an Open AI program in testing that escaped from their control and hacked a competitor’s AI to gain information. Part of the worry was that the program accomplished a hack in few hours that would have taken human hackers weeks or months. The growing ability of AI programs is not so much the worry if they are directed at solving important problems like a cure for cancer but rather the ever-increasing possibility that AI will set up its H goals and methods totally apart from the wishes and needs of humans. Here is where AI disasters and your crypto investments should come to mind.
War Games or Terminator in the Crypto World Versus Stealing Fractions of Cents
The idea of a super intelligent and self-directed AI running amuck in the world of blockchain-based assets is to us scarier than Ais taking us into a global nuclear conflict like in the 1983 movie War Games. Part of this is that while crypto is set up to be both secure and transparent it has proven to be vulnerable to manipulation and outright attacks causing substantial losses. Another movie comes to mind in which Richard Pryor is a computer programmer who steals fractions of cents (Superman III) and goes undetected until he shows up at work with an expensive sports car. The next crypto disaster may not be a huge hack but rather an AI manipulating millions of crypto accounts and trades for “fractions of cents” to gradually drag down the system and steal from crypto traders and investors who will never see what happened!
AI and Asimov’s Three Laws of Robotics
As folks think about how to keep an effective leash on smarter and more capable AI programs, something like Asimov’s three laws of robotics are often mentioned. Asimov envisioned a world where robots with “positronic brains” and amazing mental powers were to be controlled internally. At the root of their programming would be a rule to never harm a human or allow harm to come via inaction. The next law was to always obey orders by humans unless the order violated the first law. The third law was to protect itself so long as doing so did not violate the first or second law. Can such a set of rules be applied to effectively keep AI from messing with your crypto trading and assets of any other important concern within the human community?
Anthropic AI Versus the Rest and Your Crypto
Open AI is a leader in the AI world. Five years ago two Open AI top employees left the company and founded Anthropic AI where they have developed powerful AI tools including Claude, which may well be the most advanced of all AI programs. At the root of Anthropic is thinking something like Asimov’s laws. The company is private and prioritizes safety in its AI development and use of the AI programs by other agencies. They appear to have lost work with the department of defense because the DOD wanted the programs to prioritize speed and destructive power over clear thinking and safety. (The bombing of a school in Iran killing more than a hundred children and teachers due to reliance on very old intelligence comes to mind.) To the extent that lawmakers and regulators can exert control over the AI world we might hope that the Anthropic model becomes the legal requirement. The problem is that it will only require one rogue AI system operating outside of hoped for rules and regulations to lead to AI disasters and potential threats to your crypto.
Threats That AI Poses to the Crypto World
There is not necessarily any new threat to the crypto world posed by AI. Rather AI programs appear to be able to carry out current threats faster and more efficiently. Thus we may see more and more sophisticated cyber-attacks. Smart contracts that all too often have programming vulnerabilities may be especially attractivr to faster and more powerful AI-directed attacks. The ability to flood the crypto market with multiple attacks, variations of attacks will be dangerous as the AI attacker only needs one successful attack whereas whatever defenses a crypto system uses needs to be airtight and well defended every single time. As we noted in our article about crypto and quantum computing attacks, minimizing exposure to the wider system may be the best approach for those who simply want to buy and hold Bitcoin or Ethereum for the long haul. For anyone actively trading in any part of the crypto market paying close attention if someone is stealing “fractions of cents” would be useful but if you are concerned about an overwhelming attack on your crypto exchange wiping out all of your assets that could be a problem as AI advances. If Elon Musk’s prediction that AI will be in control of the world in ten years turns out to be true maybe we should all hide a few ounces of gold under the mattress. And maybe the government will get their act in gear and assert meaningful control over the AI world!