Wolfram|Alpha is now open for business. It has lofty aims:
Wolfram|Alpha's long-term goal is to make all systematic knowledge immediately computable and accessible to everyone. We aim to collect and curate all objective data; implement every known model, method, and algorithm; and make it possible to compute whatever can be computed about anything.
So I thought I'd throw a few questions at it that reflect some of my current or recent interests. It failed almost completely. Here are the questions. Except when otherwise noted,
every one of these questions produces the response "Wolfram|Alpha isn't sure what to do with your input."
- compare salaries paid to males to salaries paid to females
- do men earn more than women
- salaries of doctors (this gets a graph of doctor salaries)
- salaries of male doctors
- how much do female doctors earn
- salaries of doctors by gender
- Salaries of nurses (this gets a graph of nurse salaries)
- amount nurses earn
OK, it isn't up to speed on gender-income issues. And it can't translate "amount X earn" into "income of X". Ask Jeeves it ain't. How about more general sociological stats? I asked it
- life expectancy in different countries
It didn't know what to do with that input, but it did suggest I ask it about "life expectancy" so I did. And got—a truncated list by country (what I'd asked for in the first place) but omitting the US and other middle-ranked places. By contrast, wikipedia lists all countries twice, first as sourced from the CIA fact book and second from a UN list. It looks as if Wolfram|Alpha has used the CIA list, as it has Macau at the top, which it is with the CIA but not the UN list. Wikipedia then gives an exhaustive link-list to rankings of countries by dozens of other metrics.
Then I asked it
- Rate of infant mortality by nation
Easily cribbed from the CIA factbook but it didn't know.
This produced a short list of insurance companies headed by Costco!
- graph heart disease by age
- heart disease versus age
No result for either of these. Turning to astronomy, I asked,
- tell about interstellar dust grains
- cosmic dust
Wolfram knew nothing. N.B. Wikipedia redirects "interstellar dust" to "cosmic dust" and has an interesting article.
- nearest stars with planets
It offered "nearest stars" as a suggestion and this produced a plain-text list of star names without distances or other data. The list ended in an ellipsis and the message "Computation timed out." Note that Wikipedia has an exhaustive list of the 100 nearest stars with distance, stellar class, R.A. and Dec., and noting which are known to have planets.
This gave a list of exactly 3 (more than 100 are known), 55 Cancri d, e, and f, and again "Computation timed out." Under the same heading wikipedia has a lengthy article with a discussion of discovery methods and a table of interesting discoveries.
In short, Wolfram|Alpha is not simply distant from its lofty goals, it is ridiculously, laughably distant from them. Perhaps it answers questions in some domains adequately. Perhaps for some areas of knowledge it actually offers more than you can get by entering the same string in the Wikipedia search box. But I haven't seen any.
No comments:
Post a Comment