Google Blogoscoped

Tuesday, January 1, 2008

The Virtual Experts Room

A great way to query knowledge other than a straightforward web search would be to have a “virtual experts room.” If you’ve been watching Star Trek you might be familiar with the concept: the room is a fully digital, holographic and artificial-intelligence-run representation of reality, or a reality that could exist but doesn’t. Then, when you have a question on any topic of choice – say, you want to find out about the technology, history and political implications of the H-bomb – you query for “h-bomb.” This will fill the room with half a dozen experts.

Search engine results of today, like those by Google, are not too far away from the concept of the virtual experts room. A good search result already lists the diverse arguments and opinions by experts. But this can be expanded on by creating artifical representations of human lookalikes.

Now, it’s not really necessary to sit with the experts in a holographic room though; small speaking avatars shown on your computer screen will do as well. A label below each avatar will read their name and job title. In the case of “H-bomb,” perhaps you’ll see a couple of guys, some in old-fashioned clothing and hair styles because they’re not from your decade, with labels reading “Pat Frank, author of ’How to survive the H-bomb’” and “Robert M. Gates, current US secretary of defense” and “Harry Truman (1884 - 1972), US President who ordered the bombing of Hiroshima and Nagasaki” and “Unnamed mother, who died in Hiroshima” and “Robert Oppenheimer (1904 - 1967), director of nuclear weapons development project” and “Bob Dylan, anti-war song writer.”

Once you set-up the room with your topic of choice, you can ask any question at all. Like: “Who invented the H-bomb?” This might yield a very quick answer, depending on whether the point is debated by the experts or not. You can also ask: “What is the exact purpose of the Polysterene foam layer in Teller–Ulam bomb configurations?” You can also ask an expert: “Tell me a little bit about yourself.” You can also start more complex discussions like: “Is utilizing H-bombs for the better of the world, all in all?” A question like this would get one expert to interact with another, potentially for quite some time until you – as virtual room owner – tell them to stop, or to refocus, or you decide to un-invite a certain member of the room.

Sometimes your question may also require a new expert to be entering the room dynamically. For instance, when in the H-bomb topic room you ask a question like “Which types of bombs will be used in 2050?”, then a futurologist may enter. And if you ask, “Are there any good jokes about bombs?”, a comedian may enter the room and start getting creative. A highly ambiguous question like “Why?”, for instance, might first trigger the expert avatars to make you specify your question, and if you decide you don’t want to, spawn a philosopher to enter the room.

But an experts room may not be restricted to just human lookalikes or clones of existing humans. An avatar could also represent a country (like “Spain”), or a type of technology (like “car”), or a time period (like “Renaissance”), and so on. You can ask the country “Spain”: When were you created? What is your highest mountain? How many tourists visit you each day? You can ask the technology “car”: Who invented you? What is the historic origin of your name? Who are the different companies building you? You could ask the time period “renaissance”: Who were your most famous artists? What started you and what defined you? What ended you and who succeeded you?

The “car” object would look like a speaking car, perhaps; the avatar representing a concept, like “renaissance,” might be shaped like a human wearing typical renaissance clothing. You could also go ahead and invite two competing technologies, countries, concepts and so on to debate with each other – like a solar roof top and a nuclear plant debating the topic “eco-friendliness" for you.

(The most horrible past attempt at these kind of information agents may be stemming from Microsoft, creators of the useless Windows search dog which first appeared in Microsoft Bob, or the annoying office paper clip. These avatars had one thing in common which any good virtual experts room would need to overcome; they were all stupid, hence all they did is represent stupidity. In fact, by giving a face to stupidity, they were able to channel our negative feelings more than a mere faceless application might: people started to hate Clippy.)

In the end, these expert avatars are all virtual, dynamic representatives of knowledge, indexed by the virtual experts room provider (like Google, Yahoo, or a company yet to be created, as the technology must be “sufficiently advanced” to be “indistinguishable from magic,” as the third of Arthur C. Clarke’s three laws puts it). They make the knowledge easy for us to grasp and they give it a human face. True to the concept of the “holodeck” from science-fiction, a real-looking debate could help get the concepts across, and help you remember them, by switching its scenery, too. A debate on the H-bomb would take place in a physics laboratory, perhaps. Furthermore, experts would all be able to utilize a chalkboard in the middle of the room, sketching their ideas as they go along. You can now associate each piece of information with both a general setting (the lab) as well as a face (like that of Mr. Oppenheimer), which may aid your memory when you want to reflect on this topic. If you do forget something, a gallery of snapshots from past rooms and the experts within them is saved for your retrieval and play-back at any point in the future.

As time goes by, you will find the expertise of some of the representatives of knowledge especially helpful. Thus, you may decide to select a couple of past avatars – say, Albert Einstein, Dwight D. Eisenhower, Marissa Mayer, Bob Dylan, Noam Chomsky, Leonardo Da Vinci, Truman Capote, Joan of Arc – to discuss a current event, or the general news of the day in a non-topic room, perhaps represented by your living room. All these experts would also be able to access past knowledge they learned through interactions with other avatars from the living room. If Bob Dylan learns a bit from Eisenhower, and vice versa, you don’t want to necessarily reset that knowledge; discussions and view points can evolve because people learn.

When inviting experts over for a “free style chat” you should keep in mind that the more the subject leaves the person’s area of expertise, the more fuzzy the AI representing the person will become; that’s because less facts about the private life of the person (which would aid in extrapolating the person’s arguments on a given subject) are publicly available for the crawlers feeding the AI with information. While an AI may research in great detail, for instance, what Albert Einstein might say about certain matters in the realms of science, querying him about his favorite ice cream flavor, or good names for a girl baby, may result in less correct data.

If you do not mind about privacy, you may also offer your own persona to the virtual experts service. Maybe you’re a guru on scuba diving. You write a blog, you say a lot on the subject in a group, you email a lot about this, you chat a lot about this. Provided a single service provider accesses all of your statements from these sources (think Google Inc, which owns Blogger, Google Groups, Gmail, Google Talk and so on), you can opt-in to let others invite you into a room. If Mary wants to go scuba diving this vacation and she stumbled upon your scuba diving blog, she can now invite you into an experts room to grill you with her questions. It would be a living FAQ (a dynamic representation of the frequently asked questions to you on the subject of scuba diving), and true to the spirit of an FAQ, if the question cannot be answered in this automated way, then Mary may still decide to drop you an email – it would merely keep your workload down, if you get a lot of questions on the subject of scuba diving.

To help the virtual room experts service better represent you, you can also let it pick (or we might say “crawl”) your brain, by asking you a couple of questions. You may also restrict your avatar to only these crawl session and keep all other data private; these training sesions may be more exhausting, but it also better allows you to separate public from private statements (other than setting up a dual account, that is: one for “public” emails, chats and so on, and another for private ones). Training the bot is still easier though than having to program a “chatterbot” yourself (some technology to do so exists today).

According to a recent Pew Internet report, 58% of people say that when they have a problem, they consult the internet for help. Only 45% said they sought out friends and family for advice. Maybe in the future, the gap between these two numbers will increase. Or perhaps, the difference will just cease to exist: at that point when the information contained online will start becoming our friend, the digital, AI-driven, interactive avatar.


Blog  |  Forum     more >> Archive | Feed | Google's blogs | About


This site unofficially covers Google™ and more with some rights reserved. Join our forum!