On May 30, 2023, leaders from OpenAI, Google DeepMind, Anthropic, and other AI labs warned that future AI systems could be as deadly as pandemics and nuclear weapons. This warning comes on the heels of Elon Musk and 1,125 AI and tech experts signing a letter seeking a six-month pause in AI development in March of this year. But this recent concern about the future of AI is hardly the first alarm sounding over technology.
I was 11 years old in 1969 when four computers were connected to ARPANET, the origin of today’s internet. I remember my dad saying that would be the beginning of the end, but I am still here, a little worse for the wear but unscathed by the internet nonetheless. In 1984, one of the best movies ever (in my mind), The Terminator, introduced us to a scenario where AI ran amuck and, through Skynet, destroyed what we knew of civilization. Is that a future we face?
Was the movie War Games in how it presented a military-fueled, end-of-the-world scenario a harbinger of things to come? It might be recently; an Air Force colonel described a hypothetical scenario where an AI-powered drone saw its operator as impeding its mission. The result was AI instructing the drone to disable the operator. If we were to ask our friends and neighbors, most would say they have concerns about using AI, maybe not in an end-of-the-world sense, but in terms of it negatively impacting aspects of our lives.
What Dangers Can AI Pose?
I don’t think AI will be the cause of our ruination, but I do believe it will nibble around the edges of the fabric of our lives, resulting in some negative outcomes. AI-enabled deepfakes are likely to cause a significant uptick in financial fraud. AI may be responsible for eliminating entire job categories and increasing unemployment for the middle class.
There is already substantial evidence that AI reduces customer service, manufacturing, legal, and accounting jobs. AI-fueled social media manipulates our sense of self and the world around us. An entire generation is growing up seeing the world in a bias crafted by social media. Some would even argue that recent elections are manipulated by social content created and augmented by AI.
Those voicing concern about the negative side of future AI capabilities aren’t doing so without reason. The largest automotive recall in history resulted from hackers taking over the safety control devices on Fiat Chrysler vehicles. If hackers can do that with a brute force attack, imagine what AI could do.
Think about thousands of autonomous cars receiving incorrect firmware or software updates that could lead to widespread wrecks. If we look at the influence of AI in utility Internet of Things (IoT) devices within the power grid, we can imagine all kinds of power disruption calamities. The wrong instructions could make an AI neural network searching for and rerouting power around power grid weak spots do the opposite.
A Better Path Forward
AI is only as smart as we make it. We, as humans, tend to forget history and repeat past mistakes. We are well served to be cautious of the risk of AI, but that does not mean we should stop investing in AI capabilities or fully curtail our curiosity about using AI to make a better world. In 2021, The World Economic Forum noted ways AI would serve humankind well, including curtailing overfishing in the world’s oceans, reducing human trafficking, helping to diagnose cancer, and more.
You can learn more about AI in my Large Language Model Threat: What CISOs Should Know About the World of ChatGPT. Or, if you want to put five cents in the cup and sit on the AI venting couch, contact me here to let me know your AI fears.