Dear members of Congress,
As AI continues to advance and shape the world around us, it's important to ensure the ethical development and use of AI. We should pass legislation addressing the potential risks and dangers of AI, prioritizing the safety and rights of the people, and encouraging the responsible development of such a powerful tool. Laws made regarding AI need to focus on the safety of all people. We should have some focus on AI transparency, specifically in decision-making processes. We need to make sure there aren't errors or biases in its decisions. We need to limit what decisions AI is allowed to make, and what needs to be decided by people. Unregulated AI will lead to many, many issues in our future and will change many lives to come.
Of course, there are the issues of privacy, job displacement, discrimination, and much more that worry us about AI. These issues are extreamly important to our society as a whole, but an issue most pressing to me is the danger of misuse of AI and technology. We have to make laws that specifically address the use of AI in creating weaponry, and the violation of human and civil rights to prevent harm to many. We need to ensure that there’s accountability and legal action for those who use it to do so much harm. There was a group of researchers who had created an AI that listed roughly 40,000 ways to invent new chemical weapons. They created this using an AI that was made with the intent to check for toxicity in medical drugs. It took only 6 hours for it to create this massive list. “The researchers simply adapted a machine-learning model normally used to check for toxicity in new medical drugs. Rather than predicting whether the components of a new drug could be dangerous, they made it design new toxic molecules using a generative model and a toxicity data set.” (“Civilian AI Tech Is Already Being Misused by the Bad Guys”) The concept of this in itself is dangerous, but even worse is that it isn’t hard or complicated to attain. There are many scenarios where this could go very, very wrong. We need to prevent them from happening before it is too late to do so.
There have already been reports from our own government of other countries using simple technology to harm and kill people. Non-state actors can modify low-cost and readily available technology into dangerous weapons, threatening public safety. Making laws to prevent AI from being so readily available is crucial for our safety. There have been reports of attempted drone attacks using drones called “killer bees” by a Japanese terrorist group. The earliest record we have of a successful drone attack was back in 2013, from The Shiite Muslim terror group. They dropped 2 explosives using drones onto rebel strongholds. What's going to happen when they get a hold of AI? AI that can improve their weapons and strategies significantly within a few hours? When anyone with internet access can get access to AI, we will have so many more dangerous issues on our hands.
There are very few laws limiting AI. Many of these laws were only recently in effect, or won't be in effect until 2024 or later. However, many of said laws proposed mainly focus on privacy concerns, automated decisions, preventing plagiarism in arts, and targeted advertising. These laws are also very important to our future and way of life, but none of these laws have focused on the use of AI in warfare or the use of AI in the creation of weapons. Even still, many of these laws proposed vary from state to state. We need nationwide laws for the safety of our people, and potentially the people of other countries. Doing so will help prevent a future where AI can decide who lives and dies.
Creating and passing such laws will ensure a safer future for generations to come. We need laws that clearly outline and define the parameters surrounding AI. What strictly is and is not allowed to be produced from AI. We need to address the severe consequences of violating these potential laws, discourage the use of AI for nefarious reasons, and to hold whatever parties that do violate these laws accountable for their actions. While there are many benefits to AI, the unregulated potential dangers are much more important than the help it can give.
Thank you for your time,
Eda Bateman