AI won’t end the world

Most are familiar with ChatGPT — a language model with uncanny, human-like qualities. Unsurprisingly the breakthroughs generated a wave of panic over robots taking over the world, with calls for regulation that may be doing more harm than good.  

AI angst is fueled by disproportionate assumptions about the capabilities of the technology. We have theories like Roko’s basilisk, a modern version of Pascal’s wager, which basically proposes that future artificial intelligence will be incentivized to torture those who do not further the progress of AI, blackmailing everyone to build it in the first place. To some it seems far-fetched, but to others has reportedly caused mental breakdowns — possibly one of the first documented cases of “AI grief”. Though interesting, we should orient ourselves. We like to fantasize about possible end of the world scenarios. Roko’s basilisk along with other AI theories (like Paperclip Theory) play into our romanticization of technological doom, but they lose most of their footing when we consider realistically how artificial intelligence will take its place in our daily lives going forward. 

Realistically, humans are far too fearful to allow an AI takeover. We already see in Europe a push for regulation by lawmakers to control its development so that it is “human centric, safe and trustworthy.” It’s safe to say the European Parliament is jumping the gun and curbing the development of a technology where we haven’t fully discovered its potential. Chatbots are the ones to blame here, even though the developments using the technology in surveillance and consumer data are more alarming. Especially in China, technological development is heavily backed by the government. 

So, we have lawmakers triumphantly suppressing the AI take over. But freezing AI development probably won’t solve much. Research will move to somewhere where it’s legal, presenting a likely worse scenario of Chinese-dominated AI research. It’s better if it’s left in the public eye, where with due time our fears will wash away and we can let the technology grow into its potential which, quite frankly, won’t be as world dominating as Silicon Valley techies boast it will be. 

AI just needs a “kill switch,” which will probably be the only regulation that doesn’t repeat any other legal principles like copyright or invasion of privacy. Undoubtedly these termination features will already be developed once AI reaches the potential to do harm outside of human influences. It’s palatable to lawmakers who have trouble understanding basic technological ideas, as demonstrated in Congress’ interrogation of Google, Facebook and Twitter two years ago. Maybe a kill switch will give Capitol Hill and the public peace of mind, but regulation won’t be doing much else besides needlessly halting technological progress. 

I think AI has significant roadblocks towards replacing humans. At the end of the day it can only model human behavior, lacking the fundamentals of a “human experience.” It can only mimic us, maybe one day to an indistinguishable degree. But there are too many nuances that compose a person that we often severely undervalue. AI is an extension of ourselves, another tool we leverage to expand our domain of knowledge and understanding. We may even be able to reach the point where we can imprint someone’s entire neurology into a computer. AI will always be under the thumb and shaped by the hands of humans; will a human within a machine reject itself? Ultimately, AI presents entertaining thought experiments, but we are asking questions about technology that we do not have yet — as existential AI problems come closer to the horizon I think our questions will have much simpler responses.