OpenAI: a cautionary tale of the GPT hallucinations-to-product pipeline So far as I can tell, at least one important webpage at the OpenAI website, which ostensibly describes how to contact the company, is the product of a hallucination by their GPT-3 model. The "How can I contact support?" page describes a way to contact the company which doesn't exist
ChatGPT is a Cybersecurity Opportunity ChatGPT generated five ideas in a few seconds, and described them well. The first two conceptually overlap with my list, arguably desccribed better, and the last three are unique and interesting contributions.
A Step Toward Grand Unified Intelligence (AGI): Wolfram Alpha + ChatGPT? Stephen Wolfram has recently suggested an integration between ChatGPT (a Generative Pre-trained Transformer) and WolframAlpha could fix one of the most annoying issues with ChatGPT as a tool. When one asks ChatGPT certain simple questions with actual known numerical answers it often seems to just guess, most frustratingly, even when
In the Year 2023 Around the turn of the century Ray Kurzweil suggested that if Moore's Law held up for a little while longer it would be possible to purchase a computer chip for about $US 1000 that had roughly the same computational capacity per second as was then estimated to be. Kurzweil estimated
Wither Siri? Lots and lots of chatter among the global tech intelligentsia about ChatGPT, GitHub Copilot, and Microsoft making a $10 billion dollar investment in OpenAI. What's conspicuously missing — almost entirely so — from the global conversation about the increasing reach of artificial intelligence in recent years, is Apple. Apple was an early
Border Collies and Alien Intelligence We humans are lucky to inhabit the planet with other biological intelligences. Generally, we barely acknowledge or even ponder their presence. We understand a few of them only a little, most remain almost entirely mysterious. Some we shaped, dogs from wolves over a stretch of forty-thousand years. Dogs likely understand
AI is Coming - It's Now Urgent to Rethink Systems Security Consider these two, largely uncontroversial, propositions: 1. AI is a force multiplier. (Artificial intelligence of the limited sort which has already arrived with GPT-3 and other large language models is already a force multiplier, this issue is accute now, today, not in some future we imagine as distant where HAL
Crypto is a Fractal Ponzi Scheme My impression has been that the entire crypto industry is an infinite Fractal Ponzi Scheme™®©*. *To my knowledge nobody else has ever described it, thusly, not even Nick Weaver "Burn it with fire", Dan Olson, or Stephen Diehl "The case against crypto".
2023: Year of the Web Security Dragon New Year's Resolution 2023: secure all the web servers! Knock if off with trying to help people "fix" things that broke because part of their system supports encrypted transport, but they haven't finished the job. Stop tolerating strategic incompetence on web security. If any of y’all have a Stack
AI The Hidden Horror of the Turing Test You might not pass a Turing Test. It's rigged against you, though, so maybe don't feel too bad. A Turing test, you may recall, was proposed by Alan Turing as a way to assess an artificial intelligence system. Most of the conversation on the subject since the Turing test was
