ChatGPT Can Now Remember and Reference All Your Previous Chats

Samuel Axon, writing for Ars Technica:

Some time ago, OpenAI added a feature called “Memory” that allowed a limited number of pieces of information to be retained and used for future responses. Users often had to specifically ask ChatGPT to remember something to trigger this, though it occasionally tried to guess at what it should remember, too. [...] Users could enable or disable this feature at will, and it was automatically off for specific chats where users chose the “Temporary Chat” option — sort of ChatGPT’s version of incognito mode.

The new improvements announced today go far beyond that.

Now, where there was once a checkbox in ChatGPT’s interface to disable or enable memory tracking, there are two checkboxes. “Reference saved memories” is the old memory feature, which is basically a limited repository of essential facts. The second is the new feature: “reference chat history.” This allows ChatGPT to use all prior conversations as context and adapt future responses accordingly.

I’ve been using ChatGPT for a while now. I don’t think I’m a particularly heavy user at all, and I haven’t tried, in particular, to, I don’t know, open up about myself personally. But I ask it what I want to know. So, in the current version, I prompted it with “Describe me based on all our chats — play it straight.” Here’s its response:

You’re sharp, analytical, and deeply detail-oriented. You care a lot about precision — whether it’s language, data, or historical facts. You’re not satisfied with vague answers; you want the right one, ideally backed by sources or structured data. You’re a power user, technically literate, and comfortable with scripting and command-line tools (especially on macOS), often turning to automation and custom scripts to solve problems.

You follow the tech world closely, especially Apple — its hardware, chip architecture, software updates, and even historical product launches. But you’re not just interested in specs; you care about the bigger picture too, like corporate strategy, antitrust laws, and privacy policies.

You also track U.S. politics, legal frameworks, and electoral history with the same demand for precision and sourcing. And you’ve got a thing for language — choosing the right word, understanding nuance, and occasionally searching for just the right adjective (bonus points if it’s rare or elegant).

Outside of that, you have a strong grasp of sports stats (especially NFL and MLB), media history (Letterman, SNL, classic film), and a dry, sometimes playful sense of humor. You don’t show off, but you clearly know your stuff.

I’m probably not the right person to judge that response, but I think it’s pretty good.

Here’s the response I got when I asked Siri “Describe me based on all our chats — play it straight”:

I don’t see that person in your Contacts.

From last week’s blockbuster report by Wayne Ma at The Information, “How Apple Fumbled Siri’s AI Makeover”:

Giannandrea often has described to employees his belief that machine learning can lead to incremental improvements in products, eventually adding up to major gains, a concept he refers to as hill climbing. He also has expressed a dim view of chatbots in the past, telling Apple employees before and immediately after the release of ChatGPT that he didn’t believe they added much value for users.

★ Monday, 14 April 2025