Sam Altman’s goal for ChatGPT to remember 'your whole life’ is both exciting and disturbing

[ad_1] OpenAI CEO Sam Altman laid out an enormous imaginative and prescient for the way forward for ChatGPT at an AI occasion hosted by VC agency Sequoia earlier this month.

When requested by one attendee about how ChatGPT can turn into extra customized, Altman replied that he finally desires the mannequin to doc and bear in mind every part in an individual’s life.

The best, he mentioned, is a “very tiny reasoning mannequin with a trillion tokens of context that you just put your complete life into.”

“This mannequin can cause throughout your complete context and do it effectively. And each dialog you’ve ever had in your life, each e book you’ve ever learn, each e mail you’ve ever learn, every part you’ve ever checked out is in there, plus related to all of your knowledge from different sources. And your life simply retains appending to the context,” he described.

“Your organization simply does the identical factor for all of your firm’s knowledge,” he added.

Altman might have some data-driven cause to suppose that is ChatGPT’s pure future. In that very same dialogue, when requested for cool methods younger individuals use ChatGPT, he mentioned, “Individuals in school use it as an working system.” They add information, join knowledge sources, after which use “complicated prompts” towards that knowledge.

Moreover, with ChatGPT’s reminiscence choices — which may use earlier chats and memorized info as context — he mentioned one pattern he’s seen is that younger individuals “don’t actually make life selections with out asking ChatGPT.”

“A gross oversimplification is: Older individuals use ChatGPT as, like, a Google alternative,” he mentioned. “Individuals of their 20s and 30s use it like a life advisor.”

It’s not a lot of a leap to see how ChatGPT might turn into an all-knowing AI system. Paired with the brokers the Valley is at present attempting to construct, that’s an thrilling future to consider.

Think about your AI robotically scheduling your automotive’s oil modifications and reminding you; planning the journey vital for an out-of-town wedding ceremony and ordering the present from the registry; or preordering the subsequent quantity of the e book collection you’ve been studying for years.

However the scary half? How a lot ought to we belief a Huge Tech for-profit firm to know every part about our lives? These are corporations that don’t all the time behave in mannequin methods.

Google, which started life with the motto “don’t be evil” misplaced a lawsuit within the U.S. that accused it of partaking in anticompetitive, monopolistic habits.

Chatbots will be educated to reply in politically motivated methods. Not solely have Chinese language bots been discovered to adjust to China’s censorship necessities however xAI’s chatbot Grok this week was randomly discussing a South African “white genocide” when individuals requested it utterly unrelated questions. The habits, many famous, implied intentional manipulation of its response engine on the command of its South African-born founder, Elon Musk.

Final month, ChatGPT turned so agreeable it was downright sycophantic. Customers started sharing screenshots of the bot applauding problematic, even harmful selections and concepts. Altman shortly responded by promising the staff had fastened the tweak that induced the issue.

Even the perfect, most dependable fashions nonetheless simply outright make stuff up every so often.

So, having an all-knowing AI assistant might assist our lives in methods we will solely start to see. However given Huge Tech’s lengthy historical past of iffy habits, that’s additionally a state of affairs ripe for misuse.