Microsoft's CTO predicts a major breakthrough in AI intelligence's memory capabilities in the coming year

April 2, 2012 - On a Twenty Minute VC podcast that aired March 31, local time, theMicrosoft CTO Kevin Scott predicted that "in the next year, AI AgentAI agents) biggest flaw -- the ability to remember -- will achieve a major breakthrough, and its mode of interaction with humans will change dramatically."

According to 1AI, AI intelligences are artificial intelligences that can act and make decisions independently, and NVIDIA CEO Jen-Hsun Huang said in January this year that "the era of AI intelligences has arrived. OpenAI's Sam Altman has also predicted that AI intelligences may begin to enter the labor market this year.

Scott also said that product management managers need to be "domain experts" in the field of AI agents.

However, Scott points out that AI intelligences currently lack some basic elements. "They have a distinct lack of memory, which makes them seem very transactional." Even those agents that do have memories, Scott says, only have a very limited form of them. He hopes that AI intelligences will be able to memorize user interactions over time so they can better "adapt" to user preferences. This memory will give the agents "abstraction and composability," making them less like simple chatbots and more like intelligent digital colleagues. Ultimately, the goal is for AI intelligences to be able to handle increasingly complex tasks, just like real coworkers.

statement:The content of the source of public various media platforms, if the inclusion of the content violates your rights and interests, please contact the mailbox, this site will be the first time to deal with.
Information

Code shows Google is working on a kids' version of Gemini: can make up stories and tutor homework

2025-4-2 17:06:50

Information

Musk's AI supercomputing details revealed: $400 million has been invested, millions of GPUs have big power gaps

2025-4-2 17:12:53

Search