March 14, 2017
An article at Recode, “Watson Claims to Predict Cancer, but Who Trained It To Think,” reminds us that even the most successful AI software was trained by humans, using data collected and input by humans. We have developed high hopes for AI, expecting it to help us cure disease, make our roads safer, and put criminals behind bars, among other worthy endeavors. However, we must not overlook the datasets upon which these systems are built, and the human labor used to create them. Writer (and CEO of DaaS firm Captricity) Kuang Chen points out:
The emergence of large and highly accurate datasets have allowed deep learning to ‘train’ algorithms to recognize patterns in digital representations of sounds, images and other data that have led to remarkable breakthroughs, ones that outperform previous approaches in almost every application area. For example, self-driving cars rely on massive amounts of data collected over several years from efforts like Google’s people-powered street canvassing, which provides the ability to ‘see’ roads (and was started to power services like Google Maps). The photos we upload and collectively tag as Facebook users have led to algorithms that can ‘see’ faces. And even Google’s 411 audio directory service from a decade ago was suspected to be an effort to crowdsource data to train a computer to ‘hear’ about businesses and their locations.
Watson’s promise to help detect cancer also depends on data: decades of doctor notes containing cancer patient outcomes. However, Watson cannot read handwriting. In order to access the data trapped in the historical doctor reports, researchers must have had to employ an army of people to painstakingly type and re-type (for accuracy) the data into computers in order to train Watson.
Chen notes that more and more workers in regulated industries, like healthcare, are mining for gold in their paper archives—manually inputting the valuable data hidden among the dusty pages. That is a lot of data entry. The article closes with a call for us all to remember this caveat: when considering each new and exciting potential application of AI, ask where the training data is coming from.
Cynthia Murrell, March 14, 2017
March 13, 2017
The flow of semi-smart software publicity continues. Keep in mind that most smart software is little more than search with wrappers performing special operations.
The proud parents of Einstein and Watson announced that one another’s smart software systems have become a thing. Salesforce has scripts and numerical recipes to make it easier to figure out if a particular client really wants to drop the service. Watson brings Jeopardy type question answering and lots of data training to the festive announcement party.
I enjoyed “Salesforce Will Be Using IBM Watson to Make its Einstein AI Service Even Smarter.” The write up strikes me as somewhat closer to the realities of the tie up than the inebriated best wishes emanating from many other “real” journalists. For example, the write up asserts:
By bringing its all-important Watson service to Salesforce and Einstein customers, IBM is determined to double-down on that huge Salesforce consulting market, not compete with it.
IBM cannot “become” Salesforce. But Salesforce generates a need for services in many large companies. The idea is that Einstein does its thing to help a sales professional close a deal, and IBM Watson can do its thing to make “sense” of the content related to the company paying Salesforce for an integrated sales prospecting and closing system.
My take is that this is not much more than a co-publicity set up with the hope that the ability of Salesforce to talk about its tie up with IBM will generate sales and buzz. IBM hopes that its PR capabilities will produce some mileage for the huffing and puffing Watson “solution.”
In my opinion, IBM is turning cartwheels to get substantial, evergreen revenue from the Watson thing. But IBM may be pushing another fantasy animal into the revenue race. Quantum computing as a service is the next big thing. Now is quantum computing something one can actually use?
Nah, but the point is that revenue is not news at IBM. Quantum computing gives the IBM marketers another drum to bang. Moving in with Salesforce provides a way to sell something, anything, maybe.
Stephen E Arnold, March 13, 2017
February 21, 2017
I read a write up which might be fake news for all I know. I live in rural Kentucky and the doings of folks in a big city like Houston are mysterious and far away. Out local doctor squeezes in humans after dealing with race horses and dogs.
I read in Forbes, the capitalist tool, this story: “MD Anderson Benches IBM Watson In Setback For Artificial Intelligence In Medicine.”
The main idea is easy to grasp, even for folks like me sitting near the wood stove in Harrod’s Creek. As I understand it, IBM Watson was supposed to be helping the doctors at the número uno cancer treatment center in their quest to eradicate cancer. I assume the idea was to make more time available to physicians and other health care givers because IBM Watson would have had answers about patient treatment. IBM Watson knew the Jeopardy answers, right. Dealing with cancer-related questions seems to me to be easier: More narrow domain, more consistent terminology, smart people, etc etc.
The possibly fake news write up says:
The partnership between IBM and one of the world’s top cancer research institutions is falling apart. The project is on hold, MD Anderson confirms, and has been since late last year. MD Anderson is actively requesting bids from other contractors who might replace IBM in future efforts. And a scathing report from auditors at the University of Texas says the project cost MD Anderson more than $62 million and yet did not meet its goals.
But there is good news, or at least face saving news. I like this statement in the capitalist tool:
The report, however, states: “Results stated herein should not be interpreted as an opinion on the scientific basis or functional capabilities of the system in its current state.”
The door is not locked. Perhaps IBM Watson will once again be allowed to dine in the MD Anderson cafeteria and spark the pixels on the MD Anderson computing devices. Every smart software cloud may have a silver lining. Right?
But the project seems to be on “hold.” If the news is fake, then the project is full steam ahead, but I think the truth is closer to something like this: The users found the system like other smart software. Sort of helpful sometimes. At other times, the smart software was adding work, time, and frustration to an already high pressure, high stakes environment.
The capitalist tool ventures this observation:
The disclosure comes at an uncomfortable moment for IBM. Tomorrow, the company’s chief executive, Ginni Rometty, will make a presentation to a giant health information technology conference detailing the progress Watson has made in health care, and announcing the launch of new products for managing medical images and making sure hospitals deliver value for the money, as well as new partnerships with healthcare systems. The end of the MD Anderson collaboration looks bad.
I have zero idea what giant conference is held “tomorrow.” But I did notice this write up, which may be a coincidence: “IBM Sees Watson As a Primary Care Provider’s Assistant.” This seems similar to what IBM Watson was going to do at the MD Anderson cancer center. The write up asserts:
IBM is prepping Watson to work alongside primary care physicians and streamline processes. The company also added features to its Watson-based health cloud services.
The IBM Watson system has been enhanced too. The write up reports:
That Watson-primary care provider connection is being rolled out in Central New York in a six-county region and more than 2,000 providers. Meanwhile, Atrius Health, based in Massachusetts, will embed IBM’s cognitive computing tools inside its electronic medical records workflow for primary care providers.
This sounds good. Perhaps this is the “real” IBM Watson news. Rapid adoption and new capabilities make IBM Watson a must have in the smart health care providers arsenal of disease fighting weapons.
But there is that MD Anderson situation.
What do I make of these apparently contradictory write ups, which I assume are fake news, of course?
- IBM Watson, like other end user smart software systems, is a disappointment in actual use. Humans have to learn how to use the system and then take time to figure out which of the outputs are the ones that are likely to be useful in a particular patient’s case. Instead of saving time, the smart software adds tasks to already stretched professionals.
- The marketing and sales pressure is great. As a result, the marketers’ explanations may not match up with the engineering realities of a search-based system. When the marketers have left the building, the users learn the reality. After normal bureaucratic jabbering, the users’ dissatisfaction become too much for administrators to deal with. Hasta la vista, Sr. Watson.
- IBM, like other outfits betting on smart software, continue to repeat the cycle of belief, hyperbolic marketing, and learning about the costs and problems the smart system triggers. So why did Fast Search & Transfer’s run to fame fall off a cliff? Why is Hewlett Packard annoyed with Autonomy Software? Why did Entopia fail? Why is Lexmark’s new owners trying to exit the search with smart software business? Answer: Hope does not make an end user facing smart system generate sustainable revenues.
Because this IBM Watson news is fake. Why worry? Smart software will lift IBM to heights not experienced since the mainframe was the go to solution to computing needs. If you have a z series, you can run IBM Watson on it. Now that’s something I wish I could experience. My hunch is that none of the docs at MD Anderson will buy a z series and load up Watson because it is so darned useful. Maybe that is the “real” reality?
How does IBM get this Watson thing under control and generating money and producing happy customers? Let’s ask Watson? On the other hand, I don’t think the outputs will be too helpful.
Stephen E Arnold, February 21, 2017
February 20, 2017
It looks like the NSA is hacking computers around the world by accessing hard-drive firmware, reports Sott in their article, “Russian Researchers Discover NSA Spying and Sabotage Software Hidden in Hard Drives.” We learn that Russian security firm Kaspersky Lab found the sneaky software lurking on hard drives in 30 countries, mostly at government institutions, telecom and energy companies, nuclear research facilities, media outlets, and Islamic activist organizations. Apparently, the vast majority of hard drive brands are vulnerable to the technique. Writer Joseph Menn reports:
According to Kaspersky, the spies made a technological breakthrough by figuring out how to lodge malicious software in the obscure code called firmware that launches every time a computer is turned on. Disk drive firmware is viewed by spies and cybersecurity experts as the second-most valuable real estate on a PC for a hacker, second only to the BIOS code invoked automatically as a computer boots up. ‘The hardware will be able to infect the computer over and over,’ lead Kaspersky researcher Costin Raiu said in an interview.
Though the leaders of the still-active espionage campaign could have taken control of thousands of PCs, giving them the ability to steal files or eavesdrop on anything they wanted, the spies were selective and only established full remote control over machines belonging to the most desirable foreign targets, according to Raiu. He said Kaspersky found only a few especially high-value computers with the hard-drive infections.
Kaspersky’s reconstructions of the spying programs show that they could work in disk drives sold by more than a dozen companies, comprising essentially the entire market. They include Western Digital Corp, Seagate Technology Plc, Toshiba Corp, IBM, Micron Technology Inc and Samsung Electronics Co Ltd.”
Kaspersky did not come right out and name the NSA as the source of the spyware, but did connect it to Stuxnet, a known NSA tool. We also learn that a “former NSA employee” confirmed Kaspersky’s analysis, stating these tools are as valuable as Stuxnet.
Menn notes that this news could increase existing resistance to Western technology overseas due to security concerns. Researcher Raiu specifies that whoever created the spyware must have had access to the proprietary source code for the drives’ firmware. While Western Digital, Seagate, and Micron deny knowledge, Toshiba, Samsung, and IBM remain mum on the subject. Navigate to the article to read more details, or to view the four-minute video (scroll down a bit for that.)
Cynthia Murrell, February 20, 2017
February 19, 2017
Hey, you love mainframes. You may have some. IBMs own. Hitachi-style plug compatibles. Whatever.
Want to run some zip zip stuff on them? Now you can load Watson and get cognitive computing for your airline reservations, your government accounting, or your bank’s back office process which no one knows how to port to Goggle-style servers.
The light shined in my mind’s dark rooms when I read “IBM Brings Machine Learning To The Private Cloud.” Nestled into the article is this statement:
BM has extracted the core machine learning technology from IBM Watson and will initially make it available where much of the world’s enterprise data resides: the z System mainframe, the operational core of global organizations where billions of daily transactions are processed by banks, retailers, insurers, transportation firms and governments.
The write up makes some bold assertions; for example, “any” language, popular machine learning framework, transaction data type, and “without the cost, latency, or risk of moving data off premise.”
The write up provides a snapshot of where IBM thinks mainframes and Watson will generate revenues; specifically:
- Financial services
My thought is that each of these markets may want to reduce their dependence on mainframes and the challenges of cost control, staffing, and rapid application development “chains.”
If Watson were selling like hot cakes, why chase mainframes? Answer: More revenue. Customer demand, in my opinion, might be the wrong answer.
Stephen E Arnold, February 19, 2017
February 10, 2017
In one eight hour period I noticed these rah rah write ups about IBM Watson doing taxes. How timely? What a coincidence that these publications ran stories about yet another Watson achievement. Everything it seems except sustainable revenue.
Here are the write ups I reviewed:
- Fast Company, “H&R Block’s Watson-Powered Robots Are Here To Help With Your Taxes” stating “Block and IBM say Watson has digested 600 million “data points” from past filings to learn tips and tricks.” I bet those IRS analysts love those “tricks.”
- TechCrunch, “H&R Block Is Now Using IBM Watson to Find Tax Deductions,” stating “Beginning Sunday, February 5th, H&R Block customers will be able to interact with the new system at the company’s retail locations.” Nifty. Foot traffic for those who want H&R Block to “do” their taxes. In short, no hands on yet, right?
- New York Times, “IBM Gives Watson a New Challenge: Your Tax Return,” stating “For IBM, the collaboration with H&R Block underlines its strategy in the emerging market for artificial intelligence technology. Watson will touch consumers, but through IBM’s corporate clients.” You may have to pay to view this apparent chunk of marketing collateral. I love the “touch” thing.
You get the idea. A huge PR push for Watson, H&R Block, a promo for a super bowl commercial, and jargon about how smart Watson because it indexes text.
Revenues? Did anyone mention revenues? Cost? Did anyone mention cost? Competitive technology? Did anyone mention competitors? Editorial rigor? Are you nuts? Rigor. What’s that?
Nah. Watson. Weakly.
Stephen E Arnold, February 10, 2017
February 9, 2017
The article on ZDNet titled IBM to Use AI to Tame Big Data in Its Second African Research Lab discusses the 12th global research unit IBM has opened. This one is positioned in South Africa for data analytics and cognitive computing as applied to healthcare and urban development. Dr. Solomon Assefa, IBM’s Director of Research for Africa, mentions in the article that the lab was opened in only 18 months. He goes on,
Assefa said the facility will combine industrial research with a startup incubator, working closely with Wits’ own entrepreneur accelerator in the same innovation hub, known as the Tshimologong Precinct. Tshimologong is part of a major urban renewal project by Wits and the City of Johannesburg.
Nowhere else in the world is there an innovation hub that houses a world class research lab,” Assefa said. “One thing we agreed on from the start is that we will make the lab accessible to startups and entrepreneurs in hub.
The lab is funded by a ten-year investment program of roughly $60M and maintains an open door policy with the University of the Witswatersrand (Wits), The Department of Trade and Industry, and the Department of Science and Technology. The immediate focuses of several early applications include Cape region forest fire prevention, disease monitoring, and virtual reality.
Chelsea Kerwin, February 9, 2017
February 9, 2017
Every now and then, interest in Watson re-emerges. Forbes published a long-read recently entitled How IBM Is Building A Business Around Watson. After gaining press during Watson’s victorious Jeopardy face-off with Ken Jennings, Watson’s first commercial applications took off. IBM sold it to Memorial Sloan Kettering Cancer Center and Wellpoint to design an advisory system for its medical staff. Other medical institutions have purchased it since then. The author asserts,
Still, the potentially is undeniable. Think about how much more effective an ordinary doctor can be with Watson as an assistant. First, even before the patient enters the room, it can analyze their personal medical history, which often runs to hundreds of pages. Then, it can compare the case history with the 700,000 academic papers published every year as well as potentially millions of other patient records. All of this is, of course, beyond the capabilities of human doctors, who typically only get a few minutes to prepare for each examination. So being able to consult with Watson will be enormously helpful.
The real value is offering Watson as a service by providing its API, so that developers in organizations can develop their own applications using its technology. Over 550 partners are utilizing this currently for everything from retail to geolocation to travel agencies. Certainly, with all the hype Watson receives, we can only expect usage to grow.
Megan Feil, February 9, 2017
February 3, 2017
I like short books. But 13 pages? If you are thirsty for knowledge about IBM’s cognitive computing push, you will want to navigate to this link and download The Promise of Cognitive Computing, originally published in February 2016. Timely. My undergraduate honors essay was about five times longer than this IBM ebook.
What’s in the scholarly gem? Here’s a sampling of the topics:
- Technology transforms
- The opportunity is providing insights
- The solution: Businesses built on cognitive computing
- The opportunity for start ups
- Six real life examples of cognitive computing products
- Six steps for developing a cognitive computing product
There are two sidebars filled with useful information; for example, a definition of unstructured information and four reasons to build a cognitive computing business using Watson. There is also a link to a 30 day free trial of Watson.
Interesting. What’s happened in a year of cognitive computing? Not enough to warrant a second edition. Apparently the 19 consecutive quarters of declining revenue has blunted some of the marketing enthusiasm for ebooks.
Stephen E Arnold, February 3, 2017
January 29, 2017
I think this write up has some drops of truth in it. I wanted to check with a former MADD volunteer, but the email address wobbled and then fell against a light pole. The title was arresting: “IBM Watson Bottles ‘Holiday Spirit’ with New RUM Created Using Artificial Intelligence.” The source? The “real” news outfit the UK Mirror.
The write up explained that Watson allegedly “produces beverage based on social media posts.” I learned:
“Holiday Spirit” is claimed to be the world’s very first data-distilled rum and was created using IBM Watson. The supercomputer analyzed data from social media posts in order to produce a bespoke rum “that tastes like a holiday”. “In just six hours Watson was able to read 15 million posts on Facebook, Instagram and Twitter relating to holidays – and find the predominant emotions and concepts in those posts,” explained Joe Harrods, big data analyst and AI expert, who works closely with Watson.
The idea was that Watson guzzled 5,000 rum reviews. Then Watson demonstrated that it was in control of its faculties by “matching emotions from the reviews with ingredients.” Finally Walked a straight line to a master blender who concocted liquor, hooch, booze, or nectar that
has a subtle vanilla flavor, medium sweetness, hints of coconut and is naturally caressed with cinnamon and allspice.
So what? Here’s the results of the breathalyzer test:
“There’s no reason that this ‘taste sensation’ couldn’t be recreated for all kinds of experiences and emotions. We’ve already seen robot bartenders that can mix custom cocktails for every different punter based on their personality…
I am delighted that I have never had a drink of alcohol. I wonder if the same might be said of Watson or possibly the marketer who blended this knock out punch for artificial intelligence. What was that question? Oh, right. I remember: “Watson, when will you generate enough money to make IBM stakeholders happy.”
After 10 consecutive quarters of declining revenue, Holiday Spirit may be in short supply.
Stephen E Arnold, January 29, 2017