Making Computers Smart: A Dumb Idea?
Nova Spivak - talking about DAML, semantic web, agents - has the metaphor disappeared? There is a need for the architecture.
Barney Pell - agents will act on our behalf - people will have a relationship with their agents (Scott Banister) - conversational interfaces - persistence - agents will have desires
What do we mean by "agent"? Barney means one thing, Nova means something else. There are many different types of agents that exist. At the very extreme, they are fully autonomous. As we move the other way, they become more related to the owner. Truly autonomous agents will not happen very soon (a lot longer). The model of the web is a client server model - autonomous agents are "off on their own".
Kaliya: agents that are built from tools that will work on your behalf.
Shannon Clark: semi-autonomous agents are pretty close at the moment - with the rise of webservices and the rise of agents moving around with AJAX and structured object data.
Barney: the next step: all this great content is distributed, if anyone wants to mix up or mashup - we are making creativity happen once again. When the knowledge of what these things mean, then you (as a normal user)
Psyche - a long time, long-term project - build common-sense like an encyclopedia - DOD review of Psyche - lots of good, deep things - when you put it all together, no unifying whole. Big challenge. But, it has had some successes - and they are opening up: ResearchPsyche for people to access the information to the community.
Noah: Wikipedia (commons) versus Psyche (cathedral) - people are now working on trying to automate the extraction of "knowledge" from Wikipedia. Nova has seen a lot of holes in Psyche.
Noah: voice will not be the end-all, be-all
Barney: we will see voice augmenting other issues, where ATMs helped augment banking.
Barney is discussing that managing user-expectations on services - what the system can do, can not do. If you apply to ideal, you will be disappointed. If you apply to current, then you will be much happier.
Nova: people do not mind making mistakes, but do not like when apps make mistakes for you. Tagging, blogging - all about mistakes.
Web is a fileserver, desktop is the fileserver - Oracle wants to be the data structure. Semantic web is trying to forward RDF - RDF is a lot simpler than a relational database structure. Just a list. Nice, simple, light-weight data-structure. WinFS - Microsoft was trying to build a database, they could recover, but not right now. Maybe later. Google is certainly heavily pushing that. Especially in a mobile world - you will need remove accessible services. The question: how relevant is the desktop?
Barney: if we had a U-DB-RI, then yes we could. But until then, no. Paper on the website about this topic.
Describe something that having the WWDB?
Barney: handling travel issues is *very interesting* - complex constraints, interactive is very interesting. Web DB will not work unless...
Nova; a better form of collective intelligence. We are already highly fragments. Organizations have pockets of inteligence, but are also very stupid at times. When we are adding metadata - how reasoning can be published as knowledge is being published. As the knowledge and the expertise become virtualized, we will get higher forms of AI. The combination of large groups of human beings, with large groups of data and reasonings. Entering the era of "groupminds" - groups of people will operate more intelligently. Groups will act more intelligence that is greater than the parts which has an intelligence that is derived from the interactions of the humans and the interface with the applications. We will become more like "social insects" - the whole will have intelligence greater than the parts.
If google is the new Microsoft, will google provide the semantic web for us? How much is google the new Microsoft in your world?
Barney: google is the NEW Microsoft. But we are talking about a whole new architecture of innovation. What will the new players apply their energies to and solve new problems. Challenge to google: attraction to improve advertising revenue versus innovation that could have ahigh-risk, high-beta. Look at how google has changed the interface (not much) whereas Microsft and Ask has. Big companies have difficulties in changing at times, but do not bet against the bug guys.
Nova: google has the $$ and expertise, but culturally - they won't. They are primarily a statistic approach, whereas the semantic web is all about linguistics and meaning.
New technology: Sparkle (RDF) - will take a bit of time (1yr or 2)
No comments:
Post a Comment