chatgpt is getting humanlike memory

 

ai is about to get really personal.

by rick richardson
technology this week

as it gets better at remembering your preferences, interests and personal information, chatgpt is starting to resemble your most reliable assistant. it will even use these memories in subsequent conversations. a minor change like this might give generative ai a more human appearance and possibly open the door to general artificial intelligence (general ai), which would allow an ai brain to function more like the gray matter in your brain.

more: lenovo readies new thinkbook laptop with transparent displaybreakthrough implant uses excess blood sugar to generate electricity | one state is now america’s clean energy paradise | telehealth advances with ai-powered clinics | fcc approves superfast wi-fi tethering | what is an heic file? | bill gates behind next-generation nuclear plant | congress: tax prep companies shared private data with google, meta for years | what an a.i.-powered workforce means for accountants
goprocpa.comexclusively for pro members. log in here or 2022世界杯足球排名 today.

the restricted test was published by openai in a blog post, wherein it was explained that the purpose of the test was to evaluate chatgpt’s (both the free and chatgpt plus versions) memory of your messages across all chats.

with this upgrade, chatgpt can now remember things you specifically tell it to remember or vaguely recall fascinating details it learns along the road, like your favorite peanut butter on cinnamon raisin bagels.

one advantage of chatgpt’s memory is that it doesn’t have to start new talks from scratch. when a chatgpt has memory, it resembles a helpful assistant that is aware of your preferences, such as your preference for coffee in the morning or your dislike of scheduling meetings before 10 a.m.

according to openai, the memory will be used for subsequent prompts in practice. if you inform chatgpt that your three-year-old child enjoys giraffes, giraffe-themed birthday card ideas may come up in later ideation discussions.

instead of just repeating back to you what it remembers about your preferences and interests, chatgpt will make better use of that data to serve your needs.

an ai that can recall several chats and use that data to assist you may seem a little unsettling to some people. it’s likely for this reason that openai allows users to choose not simply to get the memories by utilizing the “temporary chat” option, which makes chatgpt appear to be experiencing some amnesia.

chatgpt allows you to go into settings and erase memories, similar to how you can remove internet history from your browser (kind of like targeted brain surgery). you can also tell chatgpt to forget something in a chat.

openai has no timeframe for when it will make chatgpt memories available to all users; as of right now, this is a test available to a select group of free and chatgpt plus users.

openai is also giving its new applike gpts memory capabilities, allowing programmers to incorporate the feature into custom chatty ais. the memories kept within the gpt will not be accessible to those developers.

it is more difficult to create an ai with long-term memory than one with a fleeting, at most, recollection of prior discussions. naturally, there are privacy implications. do you have to worry about your details showing up in other people’s chatgpt discussions if chatgpt is randomly memorizing bits and pieces of information that it deems interesting or relevant about you? most likely not. memories won’t be included in chatgpt’s training set, according to openai.

openai adds in its blog, “we’re taking steps to assess and mitigate biases, and steer chatgpt away from proactively remembering sensitive information, like your health details – unless you explicitly ask it to.” that might help but chatgpt must understand the difference between useful and sensitive info, a line that might not always be clear.

in the end, this upgrade might have important ramifications. even though chatgpt can appear fairly human during prompt-driven talks, more than a few billion neurons still separate us from its hallucinations and hazy memories of sometimes even how the conversation began.

recollections, particularly information that is casually relayed back to you during chatgpt talks, have the power to alter your perception. our shared experiences and recollections of other individuals influence our connections with them. they help us shape our conversations and encounters. it’s the way we bond. we’ll undoubtedly feel closer to a chatgpt that recalls our dislike of spicy cuisine and our passion for anything rocky balboa.

one response to “chatgpt is getting humanlike memory”

  1. roger rotolante

    by the shades of elisa!!! it make you think it is smart, by remaining you of what you told it!!!