Out on a Limb
Reading OpenAI’s recent announcement of GPT 5.5, I was struck by the following pull-quote:
One engineer at NVIDIA who had early access to the model went as far as to say: “Losing access to GPT‑5.5 feels like I’ve had a limb amputated.”
They presumably intended this to be taken as an endorsement, but I read it as a warning. I don’t want to come to rely on a tool from a single vendor to such an extent that I can’t work without it. As demonstrated recently by OpenAI’s competitor Anthropic, that vendor can unilaterally degrade the service, by accident or design, or perhaps jack up the price by five times. Given the staggering expense in serving increasingly complicated models in increasingly complicated ways, both of these seem less like if then when.
A similar criticism could be levelled at cloud providers like AWS, and many other things, and it’s always wise to approach any such entanglement with eyes open. However, AI seems like a special case, as the alternative that it’s foreclosing is not another vendor, but your own skills. Despite the myriad problems with and around LLM-based coding tools, there definitely seems to be something useful there. Leaning on them too much, though, seems like a recipe for ending up in a very uncomfortable place when the music stops.