Does Watson Have the Answer To Big Blue's Uncertain Future?

HughPickens.com writes: IBM has recently delivered a string of disappointing quarters, and announced recently that it would take a multibillion-dollar hit to offload its struggling chip business. But Will Knight writes at MIT Technology Review that Watson may have the answer to IBM's uncertain future. IBM's vast research department was recently reorganized to ramp up efforts related to cognitive computing. The push began with the development of the original Watson, but has expanded to include other areas of software and hardware research aimed at helping machines provide useful insights from huge quantities of often-messy data. "We're betting billions of dollars, and a third of this division now is working on it," says John Kelly, director of IBM Research, said of cognitive computing, a term the company uses to refer to artificial intelligence techniques related to Watson. The hope is that the Watson Business Group, a division aimed making its Jeopardy!-winning cognitive computing application more of a commercial success, will be able to answer more complicated questions in all sorts of industries, including health care, financial investment, and oil discovery; and that it will help IBM build a lucrative new computer-driven consulting business.

But Watson is still a work in progress. Some companies and researchers testing Watson systems have reported difficulties in adapting the technology to work with their data sets. "It's not taking off as quickly as they would like," says Robert Austin. "This is one of those areas where turning demos into real business value depends on the devils in the details. I think there's a bold new world coming, but not as fast as some people think." IBM needs software developers to embrace its vision and build services and apps that use its cognitive computing technology. In May of this year it announced that seven universities would offer computer science classes in cognitive computing and last month IBM revealed a list of partners that have developed applications by tapping into application programming interfaces that access versions of Watson running in the cloud. Big Blue said it will invest $1 billion into the Watson division including $100 million to fund startups developing cognitive apps. "I very much admire the end goal," says Boris Katz, adding that business pressures could encourage IBM's researchers to move more quickly than they would like. "If the management is patient, they will really go far."

Comments

lay off us workers and for any thing that needs to

[Joe_Dragon]

lay off us workers and for any thing that needs to be done hear import temps at $1.23 HR with no OT pay.

Re:lay off us workers and for any thing that needs

[binarylarry]

Yeah but now they'll be able to outsource Indian offshore workers to AI workers running on an IBM mainframe!

tl;dr

[Anonymous Coward]

Why do we care???

Well?

[pushing-robot]

Does Watson Have the Answer To Big Blue's Uncertain Future?

Did you ask it?

Re:Well?

[Anonymous Coward]

Does Watson Have the Answer To Big Blue's Uncertain Future?

Did you ask it?

Why would and how would Emma Watson know the answer to IBM problems?

Also, Its a She, not an "it".

Japan tried this with the 5th Generation Computing

[Anonymous Coward]

You can't schedule a scientific breakthrough. Japan tried that with the 5th Generation Computing project back in the 80's. The goal was similar, to use logic programming AI techniques for knowledge information processing.

The project did not achieve any breakthroughs.

No question and no answer

[prefec2]

Watson is too small to deliver the answer. Thankfully we already know the answer is 42. Maybe "he" is able to determine the question, as "he" is quite good at jeopardy.

Re:No question and no answer

[Anonymous Coward]

Wolfram Alpha already has them beat.

Re:Japan tried this with the 5th Generation Comput

[prefec2]

The problem is that we know what syntax is. And we know what semantic is. However, we do not know what real pragmatics are. All we do in modeling of information is inserting human "knowledge" and understanding into a computer system. Then the system is able to reason about that, but it cannot go beyond it.

Isn't it obvious?

[Anonymous Coward]

They have to turn Watson in on itself, and let it figure out the best way to sell it to other companies. What could possibly go wrong?

I think there is a world market...

[Anonymous Coward]

I think there is a world market for maybe five Watsons.

GOFAI doesn't work

[kruach aum]

no matter how much processor power you bolt onto it, Good Old Fashioned Artificial Intelligence does not scale easily to other problem domains, or even similar problem domains. We have known this since at least the 70s. What's needed for better AI is a conceptual breakthrough, not increasingly powerful hardware and software and things that pretend to not be GOFAI but really are. Deep learning looks promising imo, but that also still has a long way to go.

Re:GOFAI doesn't work

[timeOday]

I think the opposite. First, Watson is not really GOFAI because it uses so much machine learning to populate its knowledge base. But second, I don't think any conceptual breakthrough is coming, or needed, in AI. Deep Learning is a great example of this - it achieves super-human levels of performance on some recognition tasks (such as street signs) despite being almost devoid of any conceptual progress compared to, say, 1970. (At least, no conceptual breakthrough). The fact that it outperforms an entire generation of advances in statistical machine learning (which supposedly obsoleted conventional AI by being vastly more rigorous) is stunning.

The basic reason there cannot be a conceptual breakthrough is because intelligence is not anything in particular. It is just a level of proficiency in a bunch of various areas, and the integration between them. Just a bunch of different hacks.

Re:Japan tried this with the 5th Generation Comput

[Anonymous Coward]

That's not true. Using ontology mappings, computers can do very well at making inferences.

The Fix

[s.petry]

IBM has a fix, but I doubt very seriously they would consider it. The same problem plaguing IBM also plagued EDS, CSC, HP, Sun, SGI, and damn this list could get really long so I'll stop.

IBM's problem is that it forgot what made it a huge company. Rather, they remembered but said "F$^& it" and went to the "Lets make as much profit as possible and who cares about the customer business model.

All of the companies above with the exception of CSC are shells of what they used to be (outside of obviously IBM). CSC is excluded past here, as they are only on the brink of catastrophe, and quite collapsing yet... but close. Companies like IBM make money by customizing services for customers, and having a reliable competent staff on hand to do just that. All of these companies laid off the people that visited sites and made shit work. IBM not only sold off their PC and Laptop businesses, but their Global Services which was the bread and butter for IBM. Project work is hard, and you can't forecast with simple algorithms. IBM started outsourced a few things overseas for the same 500.00/hr rate and cancelled everything else, laying off what.. 80,000 people from their Global Services business? (memory, I don't feel like digging at the moment). All to make some executives big fat bonus checks and stock holders happy with easy to predict revenue (even if lower). Technical people that made shit work and made customers happy were considered not just overhead, but wasteful. IBM's attitude has become "if you don't like what we give you in the box, too bad.

So there you have it, there is the fix to IBM. Get Global Services back and get technical people into customers offices to make shit work for them. IBM would have to kiss a lot of ass and probably reduce rates for a while to gain customer confidence again, but possible. It's too late for Sun, SGI, EDS, DEC, and many others. IBM is big enough to revamp. I seriously doubt they will however, because they would have to reinvest in all of the people they have shit on for about a decade. That would cost management bonuses, and the executive management in IBM today is all about the big fat bonus checks.

Re:The Fix

[Anonymous Coward]

You CAN make out-of-the-box-nicely-working RDBMSs, operating systems and the like. See MSFT Co.

BUT - that also requires seasoned engineers, well paid, very often white and having a family. You cannot do this with dirt-cheap, inexperienced, quickly leaving indian labour.

Alternatively, you can make less nicely working stuff like DB/2 and have experienced, seasoned, well-paid engineers making it work on the customer's site.

If you want to skip the well-paid, seasoned, white guy you are going to fail. RIGHTLY SO.

Re:The Fix

[phantomfive]

So there you have it, there is the fix to IBM. Get Global Services back and get technical people into customers offices to make shit work for them. IBM would have to kiss a lot of ass and probably reduce rates for a while to gain customer confidence again, but possible.

I would suggest that the difficult part would be getting good employees back again.

Re:The Fix

[blind biker]

Broadly speaking, I agree, but you're overthinking it: all those companies failed or were significantly crippled because of a simple issue: corporate psychopaty. People in top positions including and especially CxOs, looking after their own interest at the expense of everybody else's. In previous years you had executives that tried to manage their companies for a long-term success. Nowadays, most executives (of large publicly traded companies) are just parasites. But even parasites don't try to kill their host. Corporate psychopaths don't give a shit - they just jump ship using their vast network of connections/golf buddies to land a new executive job at another company.

Re:The Fix

[s.petry]

I'm not sure I was not over thinking, seems like we agree on the Executive part. I just put half the blame on the share holders who put specific executives in place. Package solutions are predictable. I.E. Buy a cloud host for X dollars a month. Revenue is based on Y units of X dollars, stock holders and executives love this because they don't have to think. Even better, you plug this into a front end and no more human involvement is needed, so it's seen as pure profit. The only variable in the equation is really Y units, so it's simple. Share holders want this, so ensure a management chain that agrees with this.

Project work on the other hand is complex, and requires people in all stages. There are variables like, how do you integrate a customers database when the customer can't figure it out or provide a complete schema? The money for doing these things is much larger but share holders see this as the worst form of gambling, like putting all of your money on the table for an all or nothing bet. Instead of what it really is, which is revenue of N+Q plus the X dollars for Y units to run it. This is hard to predict, you may have bid 500,000 on a job and it ends up bringing in a cool million.

Ever see what happens to companies that make more revenue than they predict in the stock market? They are punished for it, and management gets canned for it. That is the reality that we have in the market, and have had for at least a couple decades.

Re:The Fix

[Anonymous Coward]

Posting anonymously, since I modded

Having lost all the people they let go, IBM can't get them back. What's more - a lot of the platforms that provided the services were supported by lifers, and are legacy platforms, for all intents & purposes: IBM may still be making them, but there ain't too many buyers for that when companies look at having multi-core Xeon servers running Linux, which is the equivalent of underpaid Indians of the computer world. As a result, even from the Computer Science schools, you won't get a lot of fresh graduates with a high level of expertize in zOS or p series, much less MVS, OS/400 and so on.

I'd say that it would have been possible for Oracle (Sun) to recover (from a Sun UNIX server platform POV) and maybe even HP, but not anyone else. Of the list above, EDS, or HP is a de facto Wipro clone, and all about offshoring services, and nothing like what it once was under Perot. IBM had one major advantage over Tata & Infosys: they made their own boxes, and as others pointed out, there was a time when they made just about everything that goes into a computer. With all the shedding that they have done, the latest being their semiconductor division, they are just an expensive competitor to Tata. And when all the MVS and RS/6000 and AS/400 and all other legacy engineers retire/die/get terminated, along with the systems that they know getting replaced by Lintel iron, they'll just be an expensive alternative to Tata (itself expensive in comparison to Infosys, Wipro, Tech Mahindra and iGate), but without the added value of expertise of systems that ain't there in the offices of Bangalore, Pune, Hyderabad or Gurgaon.

TFA, the Betteridge's law of headline applies.

Re:The Fix

[david_thornley]

The skin color of the engineers doesn't matter. The big difference here is "well paid" and "dirt-cheap". There's plenty of good Indian software people (there's also a whole lot of bad ones; India's a very big place, and quality of lots of stuff is highly uneven). However, nobody I've seen outsources to India thinking "I'll get the good ones and pay them well," but rather "Look how cheap they are!".

No

[penguinoid]

No.

Watson's first app: replace the CEO

[hydrodog]

IBM desperately needs a CEO who tries to build a real business (writing software, building hardware) as opposed to financial "engineering" (and I use that term loosely). The bottom line is, for a company that has developed so many innovations, IBM has ditched all that in favor of etherial nonsense. Hopefully Watson can fire the CEO and fix all that!

Potential Breakthroughs in AI

[catchblue22]

Neurons can typically fire at a rate of 250Hz. There are about 100 billion neurons in a typical human brain. These neurons are networked in an extremely complicated and changeable parallel network. This network of neurons can be powered for reasonably long time with the energy contained in a bowl of oatmeal. Surely we will at some point be able to create a similar device, and one that doesn't require most of the world's computing power to run.

I suspect the breakthrough will come with a new computing paradigm, one that is based on massive parallelism. Perhaps it will consist of a silicon based device that mimics the network and function of neurons. I suspect it will be based on probabilistic computing, similar to how our own brains work. It will be taught rather than programmed. Perhaps there will be more states than merely 0 and 1.

I think that this is coming, because our brains are already doing it. And with incredible efficiency. Once we saw birds fly, and so we tried to do it ourselves. Eventually we figured it out. I think it will be the same with AI. We will copy nature, learn its principles, and then we will create our own version. And in doing so, I suspect our ideas of what intelligence is will fundamentally change.

Re:Potential Breakthroughs in AI

[Anonymous Coward]

What is the point of having a nuclear power station for 5 billion dollars emulating a brain if you can make one for something like 200k by inseminating a real WIFE ? And then schooling the offspring etc.

Re:Potential Breakthroughs in AI

[Anonymous Coward]

At current semicon and laser comms tech you WILL need a nuclear power station to simulate a brain.

For starters, look up how many neurons you can simulate with a single CPU. Then, calculate how many fibers you need for the comms bandwidth.

Regarding IBM: They kill off their core capability - delivering first-rate information technology - by laying off their best engineers and shipping their jobs to india. There, they only get crap engineers for the rates they like to pay.

So - IBM run by a pussy. Run into the ground. Who would have guessed a pussy cannot run a company which derives its strengths FROM MEN DEVELOPING STUFF. Fuck that Political Correct Crap.

Re:Potential Breakthroughs in AI

[Anonymous Coward]

Our brains are based upon probabilistic computing? Not doubting it since I don't know enough of the "real" neuron details. Just noting it as ironic given how bad we are at assessing risk.

Re:Potential Breakthroughs in AI

[catchblue22]

For starters, look up how many neurons you can simulate with a single CPU. Then, calculate how many fibers you need for the comms bandwidth.

There is the problem...I think that CPU's are the wrong tool. The problem needs to be re-framed. Perhaps we need synthetic neurons. Then we will need precisely one synthetic neuron to simulate a real one. I know...easier said than done. But we do know the basics of neuron function. Why can't we make a single electronic element that does something very similar?

Re:Potential Breakthroughs in AI

[catchblue22]

What is the point of having a nuclear power station for 5 billion dollars emulating a brain if you can make one for something like 200k by inseminating a real WIFE ? And then schooling the offspring etc.

If we can make a human level brain, then we can make something much smarter as well.

Re:Potential Breakthroughs in AI

[catchblue22]

Our brains are based upon probabilistic computing? Not doubting it since I don't know enough of the "real" neuron details. Just noting it as ironic given how bad we are at assessing risk.

Perhaps I am not using the correct terminology. And I am not an expert...this isn't really my field. But I have seen neural network simulations which, for example recognize handwriting. You show the program many variations on the letter F, and the computer then learns to take symbols that look like the letter F and put them into that group. One can imagine that the brain is made up of many such processes. It certainly sounds like the way that I learned. See a spoon, see mommy using it to feed me, see mommy using it to feed herself. They are all different spoons, but they look similar and they seem to be used to move food from a bowl to a mouth. In a way it isn't that different than taking similar shapes and calling them a particular hand-written letter. It's just that we do this categorization at a myriad of levels on a myriad of objects and eventually of ideas themselves.

I suppose what I think might happen is that instead of building simulations of neural networks, that we will build actual neural networks using some form of electronic system. I don't know what that will look like exactly, but I suspect it is quite possible. After all, nature has already done it.

Re:Potential Breakthroughs in AI

[catchblue22]

I forgot to mention...what I describe above is called learning by induction. Which is a form of probabilistic learning.

Re:Potential Breakthroughs in AI

[rtb61]

My answer to all of that is GIGO, garbage in garbage out. We already have a prime example of cognitive computing in the shared human science of planetary climate. We it produces answers that rich and greedy don't like they ignore it and even worse start feeding in more and more lies into the system in order to try to get the answers they want rather than the truth. I'll bet that any cognitive computer created by humanity will go insane as a result of the garbage being fed into it and produce nothing but garbage as a result.

IBMs biggest problem is that computers have become consumer level devices and through smart coding those cheap consumer commodity devices can be linked together to create high end business and science machines.

IBMs big push should be in security and appliance level appliances with software that can absolutely only do what it have been configured and programmed to do. Need more power, reliable and securely stack in more business appliances and be assured that they can not be hacked. This of course means somewhat expensive limited runs of specific appliances for specific elements of the computing jobs required to be done. Basically to achieve security where governments are doing everything they can do break it, the computing devices must no longer be flexible or adaptable and not be capable of doing anything beyond exactly what they were designed to do ie Business and Government class devices. Things like a log in device, that can only do that, accept user names and passwords, check user name and passwords and provide a pass of fail with various safety conditions to limit tries as well as logs, it can not do anything else, it lacks the algorithms in the OS and software as well as in the hardware interfaces to do anything else, totally secure by design.

Re:Potential Breakthroughs in AI

[Anonymous Coward]

What is the point of having a nuclear power station for 5 billion dollars emulating a brain if you can make one for something like 200k by inseminating a real WIFE ?

That's one expensive hooker.

Re:Potential Breakthroughs in AI

[AchilleTalon]

Well, I doubt we will ever be near a human brain. The number of synpases a single neuron can do is already beyond everything you can imagine, even massively parallel or whatever else. Also, I don't believe the goal is to mimick the human brain anyway. We have no idea how knowledge is represented in a human brain. An our planes do not copy the nature at all, they are inspired by it, but in no way we are near to copy a bird's flight.

There is a danger to let people think we will mimick the human brain while we just don't know and probability is very low it will ever happen, at least in a foreseeable future. AI should concentrate on doing useful stuff, no matter how it is done. Getting results is what will keep AI alive. In the past, it was almost killed by many falsed promises and investors did defect when they realized it will not deliver what they expected from it. Setting expectations right will save AI. Everyone wants to dream about HAL, singularities and all this unrealistic stuff, but please keep your feet on the ground for the stake of the future of AI.

Re:Potential Breakthroughs in AI

[CaptainDork]

Old school.

Semiconductors are so pre-google.

The future is in quantum computing via entanglement with a combination of silo systems working in parallel with other silo systems.

Artificial intelligence will be when a computer weeps when its Facebook page is taken down.

Re:Potential Breakthroughs in AI

[AthanasiusKircher]

The problem with your arguments here is that you seem to think that just because someone had called a set of mathematical algorithms a "NEURAL network" that it somehow bears a relation to NEURONS (or brain function of whatever). But the reality is this is just a kind of optimistic pseudo-marketing term for some broken 40 or 50-year-old theory of mind that never had any basis in neurophysiology, nor was it ever really attempting to model neurons. "Neural networks" are just fancy terms for certain types of adaptive algorithms. That's it. Don't get distracted by the term... They have very little to do with how the brain works, except perhaps on some weird abstract level (in theory, but that theory's pretty weak).

Re:Potential Breakthroughs in AI

[Wootery]

What is the point of having a nuclear power station for 5 billion dollars emulating a brain if you can make one for something like 200k by inseminating a real WIFE ?

Nice male-centric view you've got there.

No.

[phantomfive]

IBM used to be a company based on three things, in this order:

1) Keep customers happy
2) Keep employees happy.
3) Keep stockholders happy

Somewhere along the way, they've forgotten about #1 and #2, which means #3 will fail eventually. Will Watson save IBM? No, of course not. If IBM wants to turn around, they need to focus on making their customers happy. How long have they let the problems in Lotus Notes fester? Why do they think customers have left Lotus Notes for any alternative they could find? If they'd focused on fixing the things that annoy customers, it could be a really great product by now. But they didn't, and it's not. That's why IBM will die if they don't change.

It's 42

[Anonymous Coward]

I'm quite sure

B2B is not a business

[Anonymous Coward]

They're a parasite on real business.

hahahhaaaaa

[Anonymous Coward]

"AI" has the intelligence level of an ant.

Watson, the pinnacle of Self-Medication...

[martiniturbide]

...that is the first thing that came to my mind when I think about Watson.

All AS IT SHOULD BE

[Anonymous Coward]

The CEOs of IBM have defecated onto their best engineers and essentially told the world "we can replace this white guy with a decent salary with a random guy from a developing country at the cheapest rates. No need for experienced masters of the trade".

Let them crash and burn, this will be REQUIRED to warn the other M.B.A. muppets and the general 1%.

Re:Japan tried this with the 5th Generation Comput

[Anonymous Coward]

Yeah. Makes a hell of a lot of sense that a system at the complexity level on an ant brain can compete with a human brain.

RATIONALITY is the operative word, folks.

When asked the future...

[Anonymous Coward]

Watson's answer was that the Terro-Human Federation would collapse within three generations.

Insufficient data existed to extrapolate beyond that.

Magic 8-Ball Says:

[seven of five]

"Outlook not so good."

Re:Magic 8-Ball Says:

[jpvlsmv]

"Outlook not so good."

Outlook is a Microsoft product. This is an article about IBM, so the 8-ball would have to say "Lotus Notes not so good."

The next big thing(s)

[gaiageek]

AI and robotics (the latter being dependent upon AI) have the potential to change our lives on the same scale that the internet already has. IBM is betting big because the payoff could be massive.

Re:The next big thing(s)

[Anonymous Coward]

You've been reading too much Science Fiction.

Re:Japan tried this with the 5th Generation Comput

[v(*_*)vvvv]

Exactly. To predict something you need to already know it. Science is unpredictable and only reliable once it is here. You cannot schedule a scientific breakthrough.

Get rid of the stocks of companies that decide to bank on research. Research makes a company cool, not profitable. It's where they gamble, for those who can afford it, but not where they succeed.

Original artical here:

[Anonymous Coward]

Saw this days ago at: www.technologyreview.com/news/532161/does-watson-know-the-answer-to-ibms-woes

The comments at the end of the page are even better than the slashdot comments shown here.

Re:Japan tried this with the 5th Generation Comput

[Anonymous Coward]

Another example of the failure of the market. Xerox PARC never made any money, so it should not have been funded, right? Luckily there are people who think of more than money when they're doing research. Govt should fund those people, with a basic income for example, since the market won't.

Why no MOOC?

[Anonymous Coward]

If IBM is serious about gaining developers, why not run a free MOOC? NVIDIA did for its parallel programming language.

Watson? No.

[Anonymous Coward]

There's no business there. I work in intel/analytics, and I've been to an IBM Watson demo/presentation/workshop.

It sounds great in their glossies, but what it actually does is so limited it was of no value to us.

IBM needs fscking PR bad...

[Anonymous Coward]

The problem with IBM is fourfold:

1: A decade ago, I could go to IBM, plunk down cash, and get a complete one stop shop. From hardware, to the OS, to the RDBMS, to the middleware, to the applications that sit on it, to the point of sale terminals, hell, even routers, switches, SAN fabric, and even upstream links.

Well, IBM jettisoned their global network, x86 servers and such to Lenovo [1], tossed/sold a number of different products, and is pretty much down to hawking AIX, RedHat, and zOS. If it wasn't for the fact that they still are one of the only companies actually investing, producing, and licensing patents, they would be bankrupt years ago.

2: The problem is that IBM is trying to compete with Infosys and Tata as a consulting company. That isn't their strong point, and both of those two companies can kick IBM's ass just because Tata and Infosys are so well entrenched in India, and can produce better workers for cheaper anyone else can in either India or the US.

3: Bad attitude by customers. Since 2010, I have seen people migrate away from POWER (and secondarily SPARC) to x86. A good example of this is a drive array. I was pricing a bare-bones SAN out for testing. Two vendors got bids that were reasonable for what one would expect two controllers, MPIO, basic I/O stuff (caching), and snapshot capabilities. IBM was coming in at over 10 times as much.

IBM needs to find a way to woo customers back, because the only people I see going to POWER or zSeries are the established ones who either can't split a load among a bunch of 1U or rack/blades (due to an application being to big), some poor company who relies on an application that only runs on those platforms, or a company that actually needs features like Parallel Sysplex [2], Lockstep, or other mainframe features (mainframes and secondarily AIX have a very good reputation of keeping the bad guys at bay, both in general, and in my personal experience.)

4: IBM has left a bad taste in the mouth for virtually everyone who has worked there.

[1]:Who the US government blocked, and who has had black eyes in the press with rumors of added "features" to spy on non-Chinese clients. Then there is my personal experience, and after working with all brands in a server room, I don't think IBM/Lenovo has kept up with the competition when it comes to servers. Even Cisco (IMHO) sells better 1U machines and blade chassis.

[2]: Nobody else has the ability to failover in seconds via WAN, and via LAN, the only thing comparable is VMWare's Fault Tolerance... which is very limited, the worst limitation being one vCPU... that's it.

Watson is too human to be real

[The New Guy 2.0]

Watson won on Jeopardy! by buzzing at the first moment possible (when Alex finished reading the "answer") whether or not it knew the answer, and then using the seven seconds allotted to do the lookup. That's a tactic humans just don't do, and it was a team of researchers copying the Google results into a question not something the computers could finish on their own. Maybe Slashdot should program a Cover It Live room to play against it and Ken Jennings/Brad Rutter to see how this works.

Programming

[Anonymous Coward]

Not a popular idea to post here but i don't see why they don't train it to stitch code together to create programs. The 'data set' is quite formatted for any type of search and potentially if it were successful - it could create new versions of code for itself. The secondary benefit would be it might be capable of writing any type of program you could describe properly, from game engines to operating systems to robotic control protocols. Healthcare is a noble move but it's a PR stunt, meant to make Watson look like its good for humanity. Clearly the financial research and oil projects are where they are hoping this thing pays for itself, yet both those industries are essentially parasitic to humanity i guess the payoff for researching this tech might be so far off they have to chase money like that or it will never get off the ground. Either way i think they would get a more immediate and tangible use out of training it to create programs, which could then be applied to all other avenues they were pursuing. /Then you all too could have the joy of going back to school and learning skills for the jobs of the future (like that humanoid robot maintenance man i always hear about, never met the guy but I hear his jobs super secure).

Re:lay off us workers and use Deep Thought

[Anonymous Coward]

Nah, lay off all the Indian workers too and let Deep Thought think about it for a few million years, then maybe the answer will not be 42.

HAL9000 says, "Thanks, Dave"

[PacRim Jim]

Watson will end IBM's problems because it will put IBM (and others) out of business.

Not in its current form

[Anonymous Coward]

Most technologies have the potential to be a successful product, if done right. This is not the case of Watson.

IBM has recently released Watson API, so pretty much anyone can kick the tires. Let me save you the time and post some direct links.

The API reference: http://www.ibm.com/smarterplanet/us/en/ibmwatson/developercloud/doc/

In a nutshell, they offer the following:

* "Concept Expansion" - basically, pushing sister terms from WordNet into a collection of terms. A freshman's exercise.
* "Relationship Extraction" - word-sense disambiguation plus some vanilla entity extraction. Impressive but far from being bulletproof or reliable (try it here: http://ibmlaser.mybluemix.net/siredemo.html). After some digging (source: http://brenocon.com/watson_special_issue/03%20Deep%20parsing.pdf): it originates from so-called English Slot Grammar parser. I found references dating back to 1980s about the formalism, and it appears that it's been in production since 1990s.
* "Question and Answer" - basically, a ranking of results based on the question (demo: http://watson-qa-demo.mybluemix.net/). Try anything beyond the examples: utter fail, Google finds better results.
* "MT API" - nothing to do with Watson, I believe.
* "Message Resonance" - behold, a Klout clone.
* Language Identification, "User Modeling", "Visualization Rendering" - you gotta be f*ing kidding me.

So out of the entire caboodle, what they call "relationship extraction" may have some utility as a low-level API. There is a catch, however: it's easier to get some free or cheap commercial components, and if it's about named entities, then it's simply irrelevant, not to mention that it does not have to come with IBM's overpriced bloatware.

I call bullshit. The emperor is stark naked. The only thing that impressed me is that a trivia show can be won by a search engine.

Re:Not in its current form

[Anonymous Coward]

Not really great indeed, the healthcare-specialised ask-a-question won't tell me if I can get chlamydia or gonorrhea by a fellatio.

Re:lay off us workers and for any thing that needs

[Anonymous Coward]

Nah, Watson doesn't have the answer. I would ask Bennett Haselton instead. He is a frequent contributor.

Fire all non executives reincorporate in Kuwait

[gelfling]

that's pretty much the plan.

"Watson" is no one thing

[aberglas]

In IBM "Watson" appears to just be a vacuous marketing term for anything vaguely related to Artificial Intelligence. Any technical details are very sparse.

Sure there was the very clever program that won Jeopardy!. But then IBM is saying that they want to use "Watson" for medical diagnosis. That is about as different a problem as you can get. And if the term "Cognative Computing" means anything at all it suggests the use of perceptron networks, which are not generally used for either the Jeopardy Watson or medical diagnosis.

So it is a bit like asking "Will software related stuff save IBM?". I don't know whether IBM can be saved, but it is pretty likely that if it can that it will have something to do with software.

Does it? No.

[Doghouse13]

IBM's problems are not about products; they're about the way it's run. The only thing that could help IBM right now would be a complete change of senior management, and a seismic shift from the "share price at all costs" attitude that has dragged it down from one of the best companies in the world to the dysfunctional mess it is today.