You may have heard recently that we’re all going to die. Not right now, but possibly as soon as 2027 or 2028. The details are scant about how this is actually going to happen. Currently, the odds that AI poses an existential threat to human life range from near zero among skeptics to 99.9 percent, according to Dr. Roman Yampolskiy, an AI safety expert and founder of the Cyber Security Lab. Even Dario Amodei, CEO of Anthropic, has put the risk of things going ‘really, really badly’ at 25 percent. That seems … high.
There doesn’t seem to be much consensus on how this would happen or who or what would cause this. So far, my favorite answer to ‘how’ comes from Dr. Yampolskiy. To paraphrase him, asking another human being how AI superintelligence would attack humans is like asking a squirrel to imagine the threat humans pose to it. The squirrel might say that humans could chase it up a tree and bite it, because that is the limit of ‘squirrel’ intelligence.
Besides the anxiety these conversations bring, they make for a strange dynamic as organizations like mine actively train our teams and work toward greater AI integration to increase our effectiveness and productivity. As a small business owner, I have been all-in on AI technology. Skills and expertise that I could never afford to bring into my organization, technology that was financially out of reach, and the ability to offload time-consuming tasks so I can focus on what matters most, are now available to my company. Today, the opportunity to level up is limited only by my ability to understand and describe what is needed. This is our great equalizer, our giant leap forward.
But leaders like me who’ve championed AI will need to think about how we address our use and promotion of AI. Anti-AI sentiment is growing and with it comes a very reasonable question: If some of the creators of this tech are asking to be stopped, why are we still embracing it?
How do we answer that? We can say that what we use to check grammar, scan email and transcribe our meetings has nothing to do with superintelligence, or recursive self-improvement. But are those everyday uses really separate from the larger AI goal? What if we all canceled our ChatGPT subscriptions in a massive boycott? OpenAI would lose revenue, yes, but it raised $122 billion in March from institutional investors, and it’s valued at nearly $870 billion. We are well beyond consumer boycotts.
At the time of my writing, Open AI scrapped its public release of its newest model, GPT-6.1 Astra, because of issues of deception and alignment. While that is good news, it feels important to note that deception and alignment are the key risks identified with AI superintelligence. In other words, the thing we are supposed to be concerned about happening in the future manifested concretely in a model meant to be released today.
Our next AI training is next week. I may have to address the mass extinction in the room.
It’s hard to know what’s true or likely with technology that moves this fast. The whole conversation is surreal, especially within the context of work. What is known is that every aspect of our lives will be shaped by this technology, one way or another. The risk is being left behind, replaced by better, faster versions of ourselves. Our goal is to thrive, which means using every tool at our disposal to keep up, and adapt to the world around us. What other choice is there? ●
Jennifer Ake-Marriott is Owner and CEO of Redmond Waltz
