The Future Is Here Today...Robots, Genetics, AI, Longevity, Singularity

wolfram-alpha1On March 5, 2009 computer scientist and theorist Stephen Wolfram struck a nerve with the search and information industry as he announced to the world the imminent launch of his new question answering service, Wolfram Alpha.   Scheduled to go live in May amidst rampant speculation and anticipation, the new service will attempt to take almost any natural language question and compute a factual answer within one or two seconds.  Now, in what appears to be his first interview since the annoucement, Wolfram opens up about his new project in a story published yesterday at H+ Magazine.

So what new information can we glean from the H+ story?  The story gives detailed information about Stephen Wolfram’s background and his own personal thoughts about his new creation.  Beyond this, however, our greatest insight came from a description of how Wolfram Alpha performs in response to three different types of questions.  From the story:

If you enter the query, “3/26/2009 + 90 days” you’ll get a page that gives a date ninety days later than the first date.   If you enter “mt. everest height length of golden gate” you’ll get a page expressing the height of Mount Everest as a multiple of the length of the Golden Gate Bridge.  If you enter “temperature in los gatos,” you’ll get something like the current temperature, a graph of the temperatures over the last week with projections for the next few days, and a graph of the temperatures over the last year.

As we can see, Wolfram Alpha will be capable of giving more than simple one word or one phrase answers.  The first question “3/26/2009 + 90 days” is a simple question that requires a concrete factual answer.  The “temperature in los gatos” question, on the other hand, is more vague and opens the door for Wolfram Alpha to provide multiple formats, such as graphs and images, to answer the question.

It seems to me that the real strength of Wolfram Alpha will be its ability to answer simple questions that have a true single, static answer.  Anything else is likely to be answered by Google better.  Case in point: take the two questions “3/26/2009 + 90 days” and “temperature in los gatos” and type them into Google.  You will see that Google does absolutely terrible with the first question (suggesting information about a man who gets 90 days in jail), but with the temperature question I find it hard to believe that Wolfram Alpha could do any better than Google has.  Unlike Google, Wolfram Alpha is not a realtime database that is constantly scouring the world for new, up to date information.  For questions that have static answers such as “3/26/2009 + 90 days” Wolfram alpha will do quite well, but for questions such as “temperature in Los Gatos” that require more dynamic answers, Google’s access to comprehensive realtime data is unbeatable.

It really gets down once again to what Wolfram Alpha is all about: it is not a Google killer or even a Google competitor.  Wolfram Alpha is after an as yet unfilled niche in the information industry of answering simple, factual questions that we generally view as having one right answer that rarely or ever changes.  Rather than compete with Google, Wolfram Alpha appears to complement Google.  In the future we can see Google either attaching a similar technology to its own service, or perhaps incorporating Wolfram Alpha itself directly into its own service as an add on.

As mentioned in our previous review of Wolfram Alpha, perhaps its greatest strength won’t even be its Google-like web interface.  Instead we can envision a simple yet powerful question answering service being integrated into thousands of devices through an open api.

Wolfram Alpha is an exciting development and the Hub will be following the service closely as its May launch approaches.  Some will say this is an over hyped idea and that Wolfram Alpha is no big deal, but we beg to differ.  We close this story with a quote from our previous coverage of Wolfram Alpha that explains our fascination with the service:

Whether intentionally or not, Wolfram Alpha has suddenly brought the idea of an accurate, computerized question answering service into the consciousness of the information industry.  It does not matter whether Wolfram Alpha’s technology is the greatest thing since sliced bread or a complete flop.  What matters is that an idea that has been around for a long time has finally come of age.  Now the race is officially on to make it a reality.

Tags: , ,

   Share    

  • j
    Both of the people who interviewed him and reviewed software on seem to be members of his fan club, so it's hard to trust their reviews. Novak is practically a nutcase as far as I'm concerned, and both Novak and the author of this new article start out by fawning over New Kind of Science. So I'm skeptical about both of these reviews.

    But you made a really good point. Even if this software totally sucks, it'll get people to think more about question answering services. I find google calculator's ability to spit out physical constants and do unit conversions is very useful, and I bet we'll see extensions of that. The field of machine learning has doing some amazing things...(photosynth, google translate are a couple examples) so learning general _knowledge_ might be something that we'll start to see.
  • MZ
    "In the future we can see Google either attaching a similar technology to its own service, or perhaps incorporating Wolfram Alpha itself directly into its own service as an add on."

    Or buying Wolfram Alpha.
  • A lot of people dissed Google when it first appeared on the scene. Now it's the throne room of the internet. Until Google, you had the cognoscenti of the internet managing who you could find and how. Google let the internet find itself.

    Wolfram Alpha aims to have the internet answer questions with computed factuality. As you say, it is only a matter of time before this or someone else does it.

    The internet, like it or not, IS an representation of what humanity thinks.
  • If this had the right APIs around it, I would love to stream information into my 3D educational games for kids. I could just write functions to connect objects and the rest could be computed in real time. I am experimenting with 3D alphabet puzzle type game but imagine if I could just compute the answers for objects, like what is the letter of the "English" alphabet that comes after A. If I wanted to make the game a Spanish game then I insert "Spanish" and the game gets NLS support. The possibilities for education are staggering. Wolfram's projects are enables for disruptive change. Keep up the great work.
blog comments powered by Disqus