Human social interactions depend on the ability to infer others' unspoken intentions, emotions, and beliefs--a cognitive skill grounded in the psychological concept of Theory of Mind (ToM). While large language models (LLMs) excel in semantic understanding tasks, they struggle with the ambiguity and contextual nuance inherent in human communication.

Human social interactions depend on the ability to infer others' unspoken intentions, emotions, and beliefs-a cognitive skill grounded in the psychological concept of Theory of Mind (ToM). While large language models (LLMs) excel in semantic understanding tasks, they struggle with the ambiguity and contextual nuance inherent in human communication. To bridge this gap, we introduce MetaMind, a multi-agent framework inspired by psychological theories of metacognition, designed to emulate human-like social reasoning. MetaMind decomposes social understanding into three collaborative stages: (1) a Theory-of-Mind Agent generates hypotheses about user mental states (e.g., intent, emotion), (2) a Moral Agent refines these hypotheses using cultural norms and ethical constraints, and (3) a Response Agent generates contextually appropriate responses while validating alignment with inferred intent. Our framework achieves state-of-the-art performance across three challenging benchmarks, with 35.7% improvement in real-world social scenarios and 6.2% gain in ToM reasoning. Notably, it enables LLMs to match human-level performance on key ToM tasks for the first time. Ablation studies confirm the necessity of all components, which showcase the framework's ability to balance contextual plausibility, social appropriateness, and user adaptation. This work advances AI systems toward human-like social intelligence, with applications in empathetic dialogue and culturally sensitive interactions. Code is available at https://github.com/XMZhangAI/MetaMind.

Richard Socher is an adjunct professor at the Stanford Computer Science Department where he obtained his PhD working on deep learning with Chris Manning and Andrew Ng. He won the best Stanford CS PhD thesis award. He is now Chief Scientist at Salesforce where he leads the company's research efforts in artificial intelligence. He previously founded MetaMind, a deep learning AI platform that analyzes, labels and makes predictions on image and text data. Richard Socher is Chief Scientist at Salesforce and an adjunct professor at the Stanford Computer Science Department.

He's the founder of MetaMind, an artificial intelligence (AI) startup that raised more than $8 million in venture capital backing from Khosla Ventures and others before being acquired by Salesforce in 2016, and he previously served as adjunct professor at Stanford's computer science department, where he also received his Ph.D. (He earned his bachelor's degree at Leipzig University and his master's at Saarland University.) In 2007, Socher was part of the team that won first place in the semantic robot vision challenge. And he was instrumental in assembling ImageNet, a publicly available database of annotated images used to test, train, and validate computer vision models. Socher -- who's now Saleforce's chief data scientist -- has long been attracted to the field of natural language processing, a subfield of computer science concerned with interactions between computers and human languages. His dissertation demonstrated that deep learning -- layered mathematical functions loosely modeled on neurons in the human brain -- could solve several different natural language processing tasks simultaneously, obviating the need to develop multiple models.

The driver in a car accident takes a picture of the damaged vehicle and sends it to an insurer for a coverage quote on the spot. A hat retailer uses data analytics to tweak its marketing formula and more than 60 percent of recipients suddenly open their messages in an email campaign. A hotel guest checks in and issues voice commands to an in-room personal assistant, ordering a rental car from the guest's preferred company that shows up outside the lobby a half-hour later. Is this the future of artificial intelligence, or is it a mad vision of computers run amok? In fact, these are all actual use cases presented during Dreamforce 2018 in San Francisco this week (pictured), and they underscore a theme that occupied much of the conversation among 170,000 attendees.

Richard Socher carries a resume that would seem to make him rather attractive to the giants of the internet. He just finished a PhD at Stanford University, where he explored a form of artificial intelligence called "deep learning," teaching machines to recognize images and understand natural language using software that operates a bit like the networks of neurons in the human brain. In recent years, the giants of net--including Google, Facebook, Microsoft, and Baidu--have seized on deep learning as a path to the future of automated computer systems, and they've been hiring researcher after researcher from the relatively small community of academics that specialize in this rather complicated technology. Socher says the big names have knocked at his door--"I had some very, very attractive offers"--but he turned them down. He wanted to start his own company, a company that would build deep learning technologies anyone can use, not just the internet giants. That company is called MetaMind, and it's backed by $8 million in funding from Saleforce.com

Amazon helped show the world how machines can learn. As far back as the late '90s, the company's online retail site would track every book, CD, and movie you purchased. As time went on, it would develop a pretty good sense of what you liked, serving up product recommendations its code predicted would catch your eye. And in the years since, the field of so-called machine learning has evolved in enormous ways, with the likes of Google, Facebook, and Microsoft training enormous networks of machines to identify faces in photos, recognize the spoken word, and instantly translate conversations from one language to another. Now, as these tech giants advance the state of the art, there's a movement afoot to bring machine learning to the business world at large.

In the consumer universe, artificial intelligence is best known as a nice-to-have if stealthy feature that can suggest movies or book rides. But AI's greater worth could well be in the money-making enterprise arena, where sales, service and marketing initiatives stand to be streamlined by the data-crunching deductive power of machine learning. That's certainly the bet customer relationship management giant Salesforce is making by unveiling Einstein, the no-brainer name given to a suite of advanced AI capabilities. With Einstein, salespeople can focus on leads that statistically show the most promise of becoming clients, and customer service reps may be better prepared to answer a rainbow of consumer queries. "The strongest aspect of Einstein is that it is deeply embedded in the platform, it's just working automatically," says Salesforce CEO Marc Benioff, whose company officially rolled out the base product Sunday.

If artificial intelligence (AI) hadn't hit the mainstream before, it did this year. Google chief executive Sundar Pichai came through with the best sound bite, saying that the world is going from being "mobile-first" to "AI-first." Apple squished AI into the iPhone, and Google stuck it in the Pixel. Facebook brought it to the News Feed, and Microsoft put it in Word. Samsung bought AI startup Viv to catch up with Apple's Siri virtual assistant.

HOW HAS ARTIFICIAL intelligence, associated with hubris and disappointment since its earliest days, suddenly become the hottest field in technology? The term was coined in a research proposal written in 1956 which suggested that significant progress could be made in getting machines to "solve the kinds of problems now reserved for humans…if a carefully selected group of scientists work on it together for a summer". That proved to be wildly overoptimistic, to say the least, and despite occasional bursts of progress, AI became known for promising much more than it could deliver. Researchers mostly ended up avoiding the term, preferring to talk instead about "expert systems" or "neural networks". The rehabilitation of "AI", and the current excitement about the field, can be traced back to 2012 and an online contest called the ImageNet Challenge.