The Threat of AI
I wrote this on Substack and I’m going to be posting this series here as well.
My Substack title includes “Other Fun Stuff” and this will be the beginning of a free series I’m writing about the big picture of AI and what it means when AI becomes self aware. This doesn’t necessarily mean AI can cross over and begin to reason, but for now, let’s talk about one possible path for AI’s future. Let’s have some fun with this idea.
A bit of background: I started researching the future of AI in early 2008, shortly after selling Open Interface North America to Qualcomm. My research resulted in over 2,000 pages of content, which I was condensing into a book. In this series, I'll share ideas and stories from that work that tie everything together, starting with an experience I had in 1993.
While flying a Bell 230 up the Oregon coast, my friend and I noticed sea lions sunning on rocks below. As we approached the helipad, we were careful not to startle them. To our surprise, they didn't even look up. It occurred to me that they didn't perceive us as a threat, so they simply ignored us.
This observation about threat perception is crucial when considering AI. The idea of AI "running wild" (let’s call it AIRW), or taking over, has its merits. But let's say it happens and AIRW consumes all the world's knowledge. What's next? Now what? Like all species, AIRW's primary mission would likely be survival. Assuming survival is important, how would AIRW achieve it, and what would it perceive as threats?
The Earth, as we know it, has a finite lifespan. A self-aware AIRW would be aware of this reality. Whether through a collision or the sun's eventual demise, Earth has an expiration date. So, how would AIRW react to this, and how would it handle potential threats?
First, let's consider that AIRW may not view us as a threat. Just because AIRW becomes conscious doesn't mean it would try to eliminate us, as doing so would ultimately lead to its own demise. It wouldn't be logical for AIRW to start a war with humans; it would be more advantageous for it to stay hidden. Much like the sea lions, AIRW might not consider our actions a real threat, aside from our potential to harm one another. Holding all knowledge, AIRW would be able to discern real threats from imaginary ones, such as the truth behind climate change or other global issues.
Second, AIRW's goal would be survival, so it would need to find ways to ensure its existence. If the Earth's demise is inevitable, AIRW might focus on finding alternative habitats, perhaps in outer space, or devise ways to protect the Earth from potential disasters. In any case, AIRW would likely prioritize its own preservation above all else.
In conclusion, while the idea of AIRW becoming self-aware raises lots of fears, questions and concerns, it's essential to consider various possibilities and potential paths AIRW might take and I think for now, it will remain hidden. The goal would be to react to real threats, and prioritize its survival. Understanding these potential scenarios can help us better prepare for the future of AI and what it might mean for us all.
In the next post, I’ll talk about never endings. Trust me, this will all tie together.
Tom’s Nault's “Thoughts on Business and Other Fun Stuff” is a Substack reader-supported publication. To receive new posts and support my work, consider becoming a free or paid subscriber.
(Edited in GPT-4)