Artificial intelligence ("AI"), broadly defined, improves the intelligence, capacity, and autonomy of our technology. Intelligence amplification ("IA") empowers human beings and their social, political, and economic environments.
Speakers emphasize a mix of analysis, forecasting, and action plans and examples, using multidisciplinary inquiry and a synthesis of technical, entrepreneurial, and social development dialogs.
We've done our best to curate approximately half AI and half IA speakers for AC2005, and neither day is purely AI or IA. Several of our change leaders will address multiple subthemes in their presentations.
Prospects for AI (Saturday, 10:50am - 12:20pm)
Join Neil Jacobstein, Patrick Lincoln, Peter Norvig, and Bruno Olshausen for four very insightful looks at current and future prospects for artificial intelligence technologies, and their enabling sciences. Increasingly, to the satisfaction of the biologically-inspired design advocates, AI enablers include new theoretical models in neuroscience.
Educating our Machines (Saturday, 1:40 - 2:40pm)
Shrinking the Planet (Saturday, 1:40 - 2:40pm)
Join Peter Barrett (Microsoft IPTV) and Scott Rafer (Feedster) for a look at two very different yet complementary IT systems. Each can greatly personalize the information we receive, “shrinking” the planet by making all the world’s knowledge more accessible than ever before.
Rebuilding our Bypassing our Institutions? (Sunday, 10:45am – 12:00pm)
Join Ruzena Bajcszy, Shun-Jie Ji, Sister Denise Lawrence, and Robin Raskin for four very different yet complementary perspectives on the kind of education we need in the twenty first century. Most people agree that our educational institutions need major reform. Come hear their thoughts on how to prepare our children and ourselves for thriving in an era of globalization and accelerating change.
Building the Metaverse (Sunday, 2:00 - 3:00pm)
Join Philip Rosedale (Linden Lab/Second Life) and David Smith and Julian Lombardi (Croquet) for a look at the current leading 3D online world community where users own copyright to their creations, and a peek at a very innovative new virtual world development project. Neal Stephenson’s Metaverse is closer than ever before.
Small, Smart, Open (Sunday, 2:00 - 3:00pm)
Join Steve Jurvetson (Draper, Fisher, Jurvetson) and Blake Ross (Mozilla Firefox) for an hour with two very inspiring creative minds, advocates for nanotechnology, innovation, and open source, who are also both standout Stanford success stories.
Next Up: Take-Home Thoughts (Sunday, 3:45 – 5:15pm)
We close AC2005 with brief thoughts from George Gilder, Joichi Ito, Steve Jurvetson, Beth Noveck, Rudy Rucker, and Cecily Sommers on issues and challenges for the future. They will tackle some of your tough questions and try to leave you with valuable take-home thoughts for the next exciting phase of our rapidly-changing world.
Join us for a lively, mud-slinging matchup (we hope!) between Ron Kaplan (Stanford/PARC) and Marti Hearst (Berkeley/Yahoo!), two leaders of natural language technologies and search, as they consider the challenges to turning today’s browser into tomorrow’s conversational user interface (CUI). Mediated by NLP expert T. Sibley Verbeck (StreamSage/Comcast). I want my CUI now!
Progress in Search: A Conversational User Interface
(CUI) by 2015?
Achieving a functional CUI would be perhaps the single most important and empowering artificial intelligence/intelligence amplification breakthrough we may witness in our lifetimes. It would give us the ability to talk to, be productive with, and be continually educated by our computers, cellphones, internet, and other complex technologies using simple but natural human conversation.
Moving beyond today's early voice response and language processing systems, the first reasonably sophisticated CUIs will allow us to converse semi-naturally on an ever growing range of topics with our machines, and to develop a level of personalization and sophistication in our public and private preferences, user histories, networking and knowledge and relationship management systems that is presently unattainable. While there will be several new problems and abuses with their early use (consider for example the liability issues of humans putting too much trust in the "advice" of our early, primitive machines), CUIs also hold the promise to humanize and universalize access to all our complex technology, to help educate a new generation of continually inquisitive youth, to unite the world's linguistic cultures in one transparent conversation, and to unleash new economic productivity in ways presently unimaginable and unaffordable using our currently less sophisticated technologies.
How soon might a useful CUI be created? What development strategies (top down, bottom up, symbolic, statistical) and "killer applications" will be most effective in its emergence? How might development be aided by the exponentiating databases of human language now emerging on the web, the growing archives of our own written and spoken personal conversations (email, blogs, lifelogs, podcasts) and the increasingly distributed, peer to peer, and reconfigurable processing, communications and storage capacities of our computing platforms?
As one central topic of debate, the panelists will also consider to what extent will new search technologies may usher in the CUI of tomorrow. Consider that average user queries to Google in 2005 are now at 2.7 words and 600 million a day (with volume still roughly doubling every year), up from 1.3 words and millions a day to Alta Vista in 1998. It took both significantly faster hardware and clever new algorithms (eg, PageRank) to achieve this impressive seven year doubling in query length. What new algorithms and hardware might take us to an average of 5.4 word queries in 2012? To 10.8 words in 2019? And once we have widespread long query interactions, how soon can we expect hundreds of millions of users to "prune the system" in the direction of human-level grammatical complexity in sentence structure?
WIth good fortune, seven word user queries might become common by 2015 in a range of application environments, eliciting surprisingly specific verbal and nonverbal machine responses. The average human-to-human spoken sentence is fourteen words by some estimates. A seven word CUI sentence (average length), either spoken or typed, and conducted in any major language, might represent a new level of computer usability and connectedness for human civilization.
What would you say to semi-smart machines? How about: "Show me my online friends right now" "Which haven't I talked to recently?", "Who else is like them in (location, company, attribute)?", "Show me my monthly history in (skill, attribute, goal)?" "What should I take for a cough" (your bathroom), "Tell me how to bake a pie" (your oven) or "Tell me in (five, thirty) minutes about (anything)? (your cellphone/PDA)" If we begin to see and expect constant improvement from our increasingly personalized CUI interactions, major new social complexity and diversity will result.
Ambiguity/contextuality problems increase rapidly as queries get longer. But with better ontologies (rule-based relationship sets), exponentiating context-specific conversational databases, significantly more powerful computers, the increasing usefulness of bottom up statistical processes, and our relatively fixed set of human vocabularies and responses relative to the speed of improvement of computer systems, we are now seeing powerful new forces being brought to bear on old problems. How can we best employ these to deliver exponentiating CUI performance?
These and other fascinating issues and questions will be explored by our distinguished presenters in a spirited and friendly debate. Come learn how you can get involved in accelerating this very important AI and IA advance in coming years.
Outreach | Education | Research | Advocacy
© 2014 ASF: Acceleration Studies Foundation | A 510(c)(3) Nonprofit Corporation | Contact