Leopold Aschenbrenner, a former OpenAI safety researcher, posted a chart on X arguing that we will have AGI by 2027 and maybe even “superintelligence” beyond 2030 thanks to millions of GPUs concentrated in 10 GW data centers. Since the launch of ChatGPT, a cottage industry of thought leaders has been stoking both excitement and fear about AGI on social media - also known as p(doom).

But if we are to take p(doom) seriously, we should first have a grown-up conversation about p(human): the possibility that we will reach human-level intelligence. I explain more in this week's newsletter: computerspeak.substack.…