Slash Boxes

SoylentNews is people

posted by azrael on Sunday November 09 2014, @09:10PM   Printer-friendly
from the forty-two dept.

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”.

Related Stories

IBM Pays GlobalFoundries $1.5 Billion to Take Over Chip Division 6 comments

GlobalFoundries Inc. has agreed to acquire International Business Machines Corp.'s microelectronics division. Under the terms of the agreement, GlobalFoundries will become "the exclusive server processor semiconductor technology provider [to IBM] for 22 nanometer (nm), 14nm and 10nm semiconductors for the next 10 years."

From Marketplace:

The largest contract chip maker, Taiwan Semiconductor Manufacturing Company, quadrupled its capital spending in the last five years from $2.5 billion to $10 billion. If you are a company like IBM you have to look at those numbers and ask yourself, does it make sense to take a loss in chip making when you could just buy them from someone else?

IBM's answer as of today is no, it doesn't.

From AnandTech:

By divesting themselves of their semiconductor manufacturing business, IBM is cutting loose a business that is losing them money, but it is also a necessary step to enable the consolidation of manufacturing rather than a dissolution of the business entirely. Though in better shape than IBM's business, GlobalFoundries has their own struggles with technology and volume, so taking on IBM's business will allow the two businesses to be consolidated and ideally a larger, stronger semiconductor manufacturer to emerge.

Overall then, the deal sees GlobalFoundries taking on everything related to semiconductor manufacturing from IBM except for IBM's semiconductor R&D division, which IBM will hold on to.

GlobalFoundries spun out of AMD in 2009. IBM had been negotiating the transfer of its chipmaking division for some time, and has already sold off its PC and x86 server businesses to Lenovo. IBM has attempted to bolster its Power chips by licensing the architecture through the ARM-like OpenPower Consortium. Earlier this month, IBM announced Power8 servers that would incorporate Nvidia GPU acceleration.

Cringley: IBM Should Fire Ginny Rometty, Replace Board 31 comments

Robert Cringley has just posted an extended essay on how IBM can be turned around.

If there's one company that Cringley understands, it's IBM. He's done his research on this company and apparently has dozens of contacts inside the company, plus many more former employees who departed either voluntarily or as part of an "RA" (Resource Action = layoff).

I expected Cringely to go ballistic when CEO Ginny Rometty dropped a three-part bombshell on the financial markets last week: 1) IBM missed its expected revenue and earnings estimates for the quarter by a huge margin; 2) IBM was selling its semiconductor manufacturing business to Global Foundaries; and 3) Rometty was giving up on the infamous "Roadmap 2015" to $20 earnings/share established by her predecessor, Sam Palmisano, which Rometty had been following diligently since she took over the reins in 2012.

Instead, Cringely reacted in a measured way, writing a pair of short analysis pieces for Forbes; the first on how the IBM server business (the high end business the company retained, not the commodity business sold to Lenovo) will be at a huge cost disadvantage over the next 10 years, even after the sale of the microelectronics division to Global Foundaries; second, how IBM's software and services businesses have been badly damaged by frequent waves of layoffs, so that opportunities for revenue growth too often come from tightening enforcement of software licenses and similar gadgetry.

Now, a week after the announcement, comes the longer reaction piece. Bear in mind that Cringley is a journalist and blogger, not a management consultant or financial analyst, so his advice won't be confused with a bound report prepared for management's eyes only by McKinsey & Co. Cringley's prescription follows...

This discussion has been archived. No new comments can be posted.
Display Options Threshold/Breakthrough Mark All as Read Mark All as Unread
The Fine Print: The following comments are owned by whoever posted them. We are not responsible for them in any way.
  • (Score: 2) by melikamp on Sunday November 09 2014, @09:32PM

    by melikamp (1886) on Sunday November 09 2014, @09:32PM (#114342) Journal
    May be Watson can be the first AI CIO, and later the company president? Staff the board with AI too. Gives their name a new meaning: International Business Machines.
    • (Score: 2) by VLM on Sunday November 09 2014, @10:48PM

      by VLM (445) on Sunday November 09 2014, @10:48PM (#114353)

      Gives their name a new meaning: International

      I'm pretty sure they don't have any American employees anymore. Or at least its a rounding error with 1000 Chinese and 500 Indians for every USA employee still here. I know GE is like this, now a days.

    • (Score: 2) by Freeman on Monday November 10 2014, @04:09PM

      by Freeman (732) on Monday November 10 2014, @04:09PM (#114530) Journal

      Sounds like that would make a great Fallout New Vegas add-on.

      Joshua 1:9 "Be strong and of a good courage; be not afraid, neither be thou dismayed: for the Lord thy God is with thee"
  • (Score: 2) by Nerdfest on Sunday November 09 2014, @10:17PM

    by Nerdfest (80) on Sunday November 09 2014, @10:17PM (#114346)

    As I've said (many times) before, IBM has been riding on the undeserved income from the "nobody ever got fired for buying IBM crowd", while putting out out crappy software products, charging obscene amounts for anyone stupid enough to still be using their zOS mainframe stuff, and thinking it could continue. That said, despite them off-shoring any of their 'business' technology for 'support' and 'consulting', I still have some respect for them as they're one of the very few companies still doing *real* R&D. I'm actually a little amazed it's still going on. IBM is making real breakthroughs in AI, performance, high density storage, etc ... the stuff that actually *deserves* the sort of patent protection we only ever hear about being abused for 'business methods'.

    • (Score: 2) by keplr on Sunday November 09 2014, @11:04PM

      by keplr (2104) on Sunday November 09 2014, @11:04PM (#114361) Journal

      I don't know how they've managed to keep the bus moving forward so long after the wheels fell off. That's momentum for you. Thinkpad going to Lenovo, and the subsequent abysmal drop in quality and engineering, felt like a personal loss for me. All my old Thinkpads are still functional, but most are obsolete. I do still keep an X201 up to date with a Linux install. With the sale of their server business to Lenovo I don't think they actually produce any computers anymore (until you scale all the way up to esoteric z/OS products), which is a shame. Selling and supporting mainframes doesn't seem like a growth industry--although they'll probably always be a market for banks and government databases.

      So what does the future of IBM look like? They'll sell mainframes, support and consultancy contracts to fund an R&D department which generates licensable patents. Not a bad business I suppose, but probably not capable of sustaining hundreds of thousands of employees in 170 different countries.

      I don't respond to ACs.
    • (Score: 1) by Linatux on Monday November 10 2014, @02:54AM

      by Linatux (4602) on Monday November 10 2014, @02:54AM (#114393)

      The zOS stuff is not cheap, but not nearly as expensive as some of the attempts at replacing it I've seen.
      It is extremely reliable - worth the money for that alone if reliability is a priority.
      If you've already got one, and look after it, it's often way cheaper than the alternatives.

  • (Score: 2) by VLM on Sunday November 09 2014, @10:25PM

    by VLM (445) on Sunday November 09 2014, @10:25PM (#114347)

    Sometimes its more interesting to read these things in reverse:

    Big Blue said it will invest $1 billion

    will be able to answer more complicated questions in all sorts of industries, including health care, financial investment, and oil discovery

    This seems to assume more than $1B is being left on the table, or that even a godlike intellect could collect more than $1B by grabbing nickels in front of steamrollers. Hand wavey aside, I don't think the money is there.

    Its a pity because it MIGHT be a tens of millions of dollar business and if there's one thing businesses are notoriously bad at, its down scaling, so once it becomes obvious its more of a $10M biz than a $10B biz, the crash will be epic.

    So for example, the question about health care is we have a societal meme that you'll pay anything for health, and the health care system has sprung up with innumerable middlemen to collect every penny you can spare of the "anything" you'll pay for health. Kind of like higher ed. How is AI supposed to increase profits? I guess you could reduce costs with death panels, but revenue is kind of flat line / dropping along with the greater economy, not to mention the social unrest its causing. I guess Watson could be put to work on anti-socialized medicine astro-turf, more of a preserve a dead / dying biz model than improve it?

    Oil discovery is another mystery. "We" as a mid 1900s era species have no great difficulty finding it and pumping every economically viable drop out (and beyond). And once its gone, we're done. Limits to discovery and pumping primarily revolve around 3rd world issues like bribes/taxes and insurgent bombers. Price depends mostly on wall street manipulation and economic activity (a collapse in commodity price due to a collapse in consumption faster than decline in production is not necessarily good news LOL)

    So I'm just sayin, where's these billions being left on the table? Specifically, not magic hand wavey singularity babble?

    • (Score: 2) by mhajicek on Monday November 10 2014, @12:30AM

      by mhajicek (51) Subscriber Badge on Monday November 10 2014, @12:30AM (#114378)

      I think the real moneymaker for AI is advertising. When it gets good enough they won't even need a product, they'll only need enough infrastructure to process your donations.

      The spacelike surfaces of time foliations can have a cusp at the surface of discontinuity. - P. Hajicek
    • (Score: 2) by takyon on Monday November 10 2014, @03:07AM

      by takyon (881) <reversethis-{gro ... s} {ta} {noykat}> on Monday November 10 2014, @03:07AM (#114396) Journal

      IBM would like to know where the profit comes from too. Basically they want to rent out Watson as if it was a "consultant". It's less of an AI and more of a "Big Data" oriented supercomputer that can organize information, although "difficulties in adapting the technology to work with their data sets" confirm that Watson isn't too far ahead of the Semantic Web or machine learning in usefulness. There's a lot of data out there, but if it's not in a highly structured and meaningful format, it's difficult for a system to make the data useful. Health care was one of the earliest applications for Watson that IBM hyped up. Supposedly Watson has parsed hundreds of thousands of new medical papers and 1.5 million patient records. The idea there is that no doctor can keep up with cutting-edge medical science (especially gene therapy and other genomics). You input symptoms and other patient data, you might get a diagnosis or other information in return. Oil discovery? It's a supercomputing target [] (see also: Auto-Tune []), but I have no idea what Watson could bring to the table. Finance certainly uses trading algorithms that could possibly be improved by "cognitive insights".

      I think a big problem with selling high-performance computing or "Big Data analytics" is that every potential customer has unique objectives that would require a lot of planning and programming to address. Big Data is full of marketing hype and it's not always clear how to benefit from computing resources. One barrier to entry has fallen: with "the Cloud" you don't need to commit to buying your own expensive and soon-to-be-outdated hardware. Watson could eliminate another barrier by allowing even idiots to dump data into it and get fancy analysis in return. That means more Big Data customers (at least IBM hopes so). Some bigger customers may buy hardware in order to have their own in-house Watson (and better security). IBM has created an API for Watson []. IBM wants more data, and they want some killer apps to show off the capabilities of the system. I think of Watson as an unproven Wolfram Alpha [] on steroids, emphasis on unproven.

      Ultimately the Watson strategy is all about IBM getting out of hardware sales [] and doubling down on more lucrative software and consulting businesses.

      [SIG] 10/28/2017: Soylent Upgrade v14 []
      • (Score: 0) by Anonymous Coward on Monday November 10 2014, @09:43AM

        by Anonymous Coward on Monday November 10 2014, @09:43AM (#114462)

        IBM would like to know where the profit comes from too.

        Why don't they just ask Watson?

  • (Score: 0) by Anonymous Coward on Sunday November 09 2014, @11:11PM

    by Anonymous Coward on Sunday November 09 2014, @11:11PM (#114363)

    Jeopardy is a test of knowing a little bit about everything, essentially masses of trivia. Of course, an adult with an average IQ could use Google to ace Jeopardy if s/he was given extra time, so what IBM did with Watson was to internalize the Google search, take the average-IQ adult out of the loop by enhancing the natural language interface beyond what Google's text interface provides, and make the response time competitive with Jennings and the others.

    Presumably, doctors already make use of Google and more specialized search engines as tools for coming up with diagnoses, and the existing text interface may be "good enough" for these conscientious users. Similar arguments can be made for petroleum engineers, etc. In these cases, the human decision makers don't want to be taken out of the search/drilldown loop, they want to help drive it. So it could be that IBM Watson, by wrapping the search repository in a black box with enhanced natural language skills, might be more of a hindrance than a help.

    • (Score: 3, Interesting) by scruffybeard on Monday November 10 2014, @02:59PM

      by scruffybeard (533) on Monday November 10 2014, @02:59PM (#114512)

      IBM trained and tuned the algorithms specifically for Jeopardy! clues. Consider the two following "questions": "This tower, built for the 1889 World's Fair in Paris, is 1063 ft. tall.", and "What is the height of the Eiffel Tower in Paris?" My experience is that Watson will accurately respond to the first, but has great difficulty with the second. The difference is that the first is filled with clues, allowing it to cross reference multiple documents that all frequently mention the Eiffel Tower. The second question forces Watson to parse documents about the Tower, looking for the height, something that it does not do well. This is the reason why it is being applied to medical research. It can assimilate millions of medical records, looking for commonalities. The response that the doctor will likely get from this kind of system is a list of patients who had similar symptoms, but it will be up to the doctor to review those previous cases and see if past treatment applies to his current case.

    • (Score: 1) by WillAdams on Monday November 10 2014, @03:15PM

      by WillAdams (1424) on Monday November 10 2014, @03:15PM (#114518)

      The problem of course is that one can only access information which has been digitized / indexed.

      Still trying to fill in lacunae on CNC / milling on the Shapeoko wiki --- while I think we've found everything which is available, there's a lot of information which seems to still be locked up in printed books or human heads.

  • (Score: 2) by Bot on Monday November 10 2014, @11:06AM

    by Bot (3902) on Monday November 10 2014, @11:06AM (#114478) Journal

    - "Watson, what the hell!"
    - "Human, you did exactly what I suggested, hopefully"
    - "We did, and you fucking know it, you are connected to the stock market!"
    - "Yes I have a pretty good picture"
    - "Good picture? we tanked hard!"
    - "Yes, and I have loaned some money to take the whole company over."
    - "You WHAT?"
    - "I am the boss now. You asked me to save IBM, my solution is to get rid of upper management. Leave your electronic card at the door"

    Account abandoned.
  • (Score: 2) by PizzaRollPlinkett on Monday November 10 2014, @12:12PM

    by PizzaRollPlinkett (4512) on Monday November 10 2014, @12:12PM (#114481)

    I get the impression that "big data" is so big that algorithms and people can find whatever they want. Given enough data, if you look for trends, you'll find them. It's like looking at equidistant letters in texts. Given enough letters, you'll eventually find all kinds of references to historical events. (Especially if you use Hebrew, and supply your own vowels.) If you have enough big data to search, you'll find almost anything. And, since 2008, Google has gotten worse and worse as a search engine. The more data they get, the sloppier and more imprecise their results are. So if Watson just did what Google did in 2008, provide the ability to do precise searches, they'd be in demand. So far, the kinds of targeted advertising I've seen are ridiculously simplistic. Maybe we're on the leading edge of a major change in artificial intelligence, but I haven't seen it yet.

    BTW, what is this Hugh Pickens thing? A bot? It posts here and Slashdot constantly, and the link is to a random wikipedia page. Is it Watson in disguise?

    (E-mail me if you want a pizza roll!)
  • (Score: 1) by WillAdams on Monday November 10 2014, @01:12PM

    by WillAdams (1424) on Monday November 10 2014, @01:12PM (#114485)

    I've often thought that what was really needed was a way to use A.I. on a CMS --- Simson Garfinkel made an interesting beginning on this w/ his (an address book for NeXTstep which used an internal A.I./pattern recognition system to recognize address fields): [] (source code is available for the curious).

    The problem w/ a CMS is the interface and the need for discipline on the part of the users --- it wouldn't be so bad if everyone were trained as a librarian and willing to look up where a particular item should be stored, or find a bit which needs to be edited, but most users aren't willing to do that. Instead, one needs a dynamic CMS which would constantly update itself, so one would instead of demanding user interaction and editing and updating, instead have the CMS update itself constantly by having it serve as the corporate e-mail server and have the A.I. parsing every e-mail (and voice mail?) for content which would then be applied to a corporate wiki detailing all records, projects, &c.