Slashdot Mirror
Microsoft Forms New AI Research Group Led By Harry Shum (techcrunch.com)
An anonymous reader quotes a report from TechCrunch: A day after announcing a new artificial intelligence partnership with IBM, Google, Facebook and Amazon, Microsoft is upping the ante within its own walls. The tech giant announced that it is creating a new AI business unit, the Microsoft AI and Research Group, which will be led by Microsoft Research EVP Harry Shum. Shum will oversee 5,000 computer scientists, engineers and others who will all be "focused on the company's AI product efforts," the company said in an announcement. The unit will be working on all aspects of AI and how it will be applied at the company, covering agents, apps, services and infrastructure. Shum has been involved in some of Microsoft's biggest product efforts at the ground level of research, including the development of its Bing search engine, as well as in its efforts in computer vision and graphics: that is a mark of where Microsoft is placing its own priority for AI in the years to come. Important to note that Microsoft Research unit will no longer be its on discrete unit -- it will be combined with this new AI effort. Research had 1,000 people in it also working on areas like quantum computing, and that will now be rolled into the bigger research and development efforts being announced today. Products that will fall under the new unit will include Information Platform, Cortana and Bing, and Ambient Computing and Robotics teams led by David Ku, Derrick Connell and Vijay Mital, respectively. The Microsoft AI and Research Group will encompass AI product engineering, basic and applied research labs, and New Experiences and Technologies (NExT), Microsoft said.
After all, Microsoft needs to develop some intelligence.
Yeah the problem of AI. The moment we figure out how to do an aspect of "AI" we call it algorithm and its no "AI" anymore.
With your logic, humans are not intelligent, as the human body is just a complex machine that follows certain rules (algorithms).
In the case of Microsoft AI, I think they could actually create a new Turing Test. If and when the AI says, "Sorry, Dave, I'm not going to install this update", then it's a sentient being.
There have been a number of philosophical critiques of artificial intelligence, the most prominent being Searle's Chinese Room argument, but there are a number of others, like Hubert Dreyfus, who have been critical and whose challenges have largely been ignored. The problems are with the fundamental concept of intelligence and how a mechanical intelligence could be constructed. First, cognitive science assumes that intelligence is a matter of a mind following certain rules in the manipulation of internal representations (Chomsky's legacy). Second, since we have machines that can follow rules and have internal representations (programs and bits) it was supposed to have been easy to move past the early successes of the field, the specialized AIs that could move blocks around, to create a general AI, but this success never came and remains elusive, while the research dogma remains unquestioned. Third, it was never an issue of the capabilities of the hardware, but the fact that human intelligence comes from a human being situated in the environment which allows for intelligent behavior, and the AI program continually rejects this approach, the mind has to be like a computer because Chomsky has to be right.
So a bunch of philosophers sit around thinking up definitions all day.... ...meanwhile the world invents computers that can drive our cars for us, crunch our data and make recommendations for us, and someday organize our economy for us.
If they define intelligence such that this doesn't qualify, then they have made the word useless. Here's a definition for you "Intelligence: whatever humans can do that computers can't." There. We will never invent AI. Happy now?