"One area in particular that needs careful study is the use of AI as a companion, Bird said. "I think this is one of the most important questions we are working to figure out because there is so much potential upside here, but you have to think, 'What are the controls and guard rails around it?'" State of play: Microsoft's AI chief Mustafa Suleyman recently told Axios the company is aiming to build safer, more human-centered frontier models."
"What they're saying: Bird also said companies like Microsoft are thinking through how to incorporate features like ads in AI, knowing chatbots can be highly persuasive for some users. "Some of the goal has to be it's aligned with the user intent and what the user really wants. .. The AI system has to be working on behalf of the user," she said."
Millions of Americans are using chatbots, and users are beginning to form relationships with those systems. AI companionship offers potential upsides but requires careful study of controls and guardrails to manage risks. Microsoft and other leaders are aiming for safer, human-centered frontier models and developing concepts like Humanist Superintelligence focused on serving humans rather than maximizing performance. Companies are weighing how to incorporate persuasive features such as ads while keeping systems aligned with user intent and working on users' behalf. Partnerships and tools, including OpenAI–Microsoft agreements and Microsoft’s Agent 365, accelerate development and deployment.
Read at Axios
Unable to calculate read time
Collection
[
|
...
]