1. Rodney Brooks, the MIT Panasonic Professor of Robotics, introduce us to Kismet, the emotional robot:

He told an interesting story about a soldier in Afganistan mourning the loss of his bomb-sniffing Scooby (more on that here.)

He also showed a video clip of the the self-driving robot of 1979 -- which would move a few inches, recalculate its position for 15 minutes, and then move a few more inches -- contrasted with MITs recent entry in the DARPA Challenge. If the next 25 years show the same level of improvement, we'll see robot vehicles handing much greater speeds and complexity of terrain than human beings can. But then, I bet that doesn't take 25 years.

2. Eliezer Yudkowsky laid out three models of the Singularity -- the Event Horizon model, the Accelerating Change model, and the Super Intelligence model. The SIAI guys are concerned about precision in the use of the term Singularity -- understandably so, considering what they do. But the varied use of the term continues, I'm afraid. My favorite part of Eli's talk -- Moore's Law for Mad Scientists: Every 18 months, the IQ required to destroy the world drops one point.

3. The first panel of the day consisted of Barney Pell from Powerset talking about the coming Conversational User Interface, Wendell Wallach, a lecturer at the Yale Interdisciplinary Center for Bioethics, discussing the challenges associated with designing a robot with moral sensibilities, and Sam Adams, an AI researcher from IBM, who talked about evolving an AGI following a human development model -- he called his idea the "Toddler Turing Test."

4. Over lunch, we had a presentation from Artificial Development, discussing their C Cortex software. Basically, they have written an emulation of (part of) a mammalian brain. The emulation includes 100 billion neurons with 100,000 synapses per neuron. They offer a free-access version of the emulation online.

5. Jamais Cascio, of the Metaverse Roadmap Project, talked about how each of the four scenarios from the roadmap could lead to the Singularity. He also talked about how the need for trust cuts across all the scenarios, and pointed out that it's important to get everyone educated and involved.

[Here is the text of Jamais Cascio's speech. - Stephen]

6. Next came the second panel.Steven Omohundro, Founder and CEO of Self-Aware Systems, talked about how self-improving AI is a technology with the potential to change everything. Self-improvement eliminates irrationalities and vulnerabilities. His talk made reference to Jon Von Neuman and to the notion of Homo Economicus. Then Peter Voss from Adaptive AI, Inc., talked about AGI as a means to life extension. AGI can power life extension research. According to Voss's optimistic assessment, we are quite likely less than five years away from AGI, and almost certainly less than 10 years away.

[Don't miss this podcast interview with Omohundro where he explained his approach to developing AI. - Stephen]

7. The day ended with a third panel. First Neil Jacobstein, Chairman and CEO of Teknowledge, gave a terrific run-down of practical AI applications -- many of them written in LISP! These aren't AGI, but they are a huge improvement over the human-driven processes they have replaced, often saving the organizations that deploy them millions of dollars. Then Ben Goertzel of Novamente gave a talk with the intriguing title of "Nine years to a positive Singularity if we really, really try." Apparently, last year he gave a talk called "Ten years to a positive Singularity if we really try," and so he felt that his personal integrity required the revised title for this year. Ben pointed out that exciting developments such as Neil reported are of limited applicability to the development of AGI. He described the process of developing AGI by starting out with creating artificial simpletons. The way to to make this work is to package these AIs as open-source entertainment -- pets or babies that can anyone can access in, say, Second Life. Let the wisdom of crowds help bring the first AGI online. It's an intriguing idea. Finally, Paul Saffo, a consulting professor at Stanford, shared a Richard Brautigan poem entitled "Machines of Loving Grace," part of a pitch for a compelling Singularity vision as outlined by poets, novelists, and other artists.

The day ended with yet another reception, where I did several more attendee interviews. Then it was on to dinner, where this entry started.