What Happens when an AI Debates Politics?

April 20, 2021

IBM machine-learning researcher Noam Slonim spent years developing a version of IBM’s Watson that he hoped could win a formal debate. The New Yorker describes his journey and the results in, “The Limits of Political Debate.” We learn of the scientist’s inspiration following Watson’s Jeopardy win and his request that the AI be given Scarlett Johansson’s voice (and why it was not). Writer Benjamin Wallace-Wells also tells us:

“The young machine learned by scanning the electronic library of LexisNexis Academic, composed of news stories and academic journal articles—a vast account of the details of human experience. One engine searched for claims, another for evidence, and two more engines characterized and sorted everything that the first two turned up. If Slonim’s team could get the design right, then, in the short amount of time that debaters are given to prepare, the machine could organize a mountain of empirical information. It could win on evidence.”

Ah, but evidence is just one part. Upon consulting with a debate champion, Slonim learned more about the very human art of argument. Wallace-Wells continues:

“Slonim realized that there were a limited number of ‘types of argumentation,’ and these were patterns that the machine would need to learn. How many? Dan Lahav, a computer scientist on the team who had also been a champion debater, estimated that there were between fifty and seventy types of argumentation that could be applied to just about every possible debate question. For I.B.M., that wasn’t so many. Slonim described the second phase of Project Debater’s education, which was somewhat handmade: Slonim’s experts wrote their own modular arguments, relying in part on the Stanford Encyclopedia of Philosophy and other texts. They were trying to train the machine to reason like a human.”

Did they succeed? That is (ahem) debatable. The system was put to the test against experienced debater Harish Natarajan in front of a human audience. See the article for the details, but in the end the human won—sort of. The audience sided with him, but the more Slonim listened to the debate the more he realized the AI had made the better case by far. Natarajan, in short, was better at manipulating his listeners.

Since this experience, Slonim has turned to using Project Debater’s algorithms to analyze arguments being made in the virtual public square. Perhaps, Wallace-Wells speculates, his efforts will grow into an “argument checker” tool much like the grammar checkers that are now common. Would this make for political debates that are more empirical and rational than the polarized arguments that now dominate the news? That would be a welcome change.

Cynthia Murrell, April 20, 2021

AI Suffers the Slings and Arrow of Outrageous Marketing

March 19, 2021

I read “Loose Lips Sink AI Ships.” Amusing. The write up begins with a sentence designed to catch my attention:

Cognitive computing is not an IBM fraud. [Emphasis added. Editor.]

Imagine. IBM and fraud in the same sentence. Even more tasty is the phrase “cognitive computing.” The phrase evokes zeros and ones which think. The implication is that smart computers are as good as a mere mortal, perhaps even better at some things.

Fraud. Hmmm.

The write up explains that one naysayer is missing the boat. The naysayer took umbrage as a marketing person’s characterization of IBM Watson artificial intelligence platform being able to “outthink human brains in areas where finding insights and connections can be difficult due to the abundance of data.”

My goodness. A marketing person exaggerating. Plus the “abundance” word evokes the image of a tsunami of information. That’s an original metaphor too.

The write up explains that AI is a whiz bang deal. The case example is Covid research. I was hoping that the author would explain how IBM Watson was lashed to a gurney and wheeled into the parking lot at a major Houston, Texas, hospital. But no. The example was Covid.

The write up explains that AI is better with bigger and faster computers. That’s good news for some companies. Also, computer reasoning is “increasing quickly.” I like increased reasoning.

There is some less than sunny news too. What a surprise. For example, neural networks are clever, not intelligent. Clever was good enough for the Google, but not enough for real AI yet. And AI systems mimic human intelligence; the systems are not quite like your next door neighbor. (I think computers are quite like my next door neighbor, but I live in rural Kentucky. That’s a consideration.)

The write up seems to strive for balance if one relates to big data, big computers, and big marketing.

Let’s ask Watson? Well, maybe not.

Stephen E Arnold, March 19, 2021

Who Should Watch Over Smart Software? No One. Self Regulation Is the Answer

March 11, 2021

I read an amusing academic paper article called “Someone to Watch Over AI and Keep It Honest – and t’s Not the Public!.” The idea is that self regulation works. Full stop. Ignoring the 737 Max event and Facebook’s legal move to get anti-trust litigation dumped, the write up reports:

Dr Bran Knowles, a senior lecturer in data science at Lancaster University, says: “I’m certain that the public are incapable of determining the trustworthiness of individual AIs… but we don’t need them to do this. It’s not their responsibility to keep AI honest.”

And what’s the smart software entity figuring prominently in the write up? Amazon, the Google, or Twitter?



The idea, at least in the construct of the cited article, is that trust is important. And whom does one trust?


How do I know there’s an element of trust required to accept this fine scholarly article?

Here’s a clue:

The paper is co-authored by John T. Richards, of IBM’s T.J. Watson Research Center, Yorktown Heights, New York.

Yep, the home of the game shown winner and arguably one of the few smart software systems to be put on a gurney and rolled out the door of a Houston, Texas medical facility.

But just in case the self regulation thing doesn’t work, the scholarly experts’ findings point to “a regulatory ecosystem.”

Yep, regulations. How’s that been working out in the last 20 years?

Why not ask IBM Watson?

Stephen E Arnold, March 11, 2021

IBM Watson and Health: Take Two Aspirin, Do Not Call Me in the Morning

March 8, 2021

IBM Watson was going to put cancer in the cupboard with AS/400 manuals. Then the billion dollar brainiac was going to deal with the Covid Rona thing. Neither worked out.

Ever since Watson blew the competition away on Jeopardy, IBM boasted that their supercomputer would enhance and/or repair industries.  The biggest mountain IBM wanted Watson to scale was healthcare and MarketScreener shares: “International Business Machines: IBM’s Retreat From Watson Highlights Broader AI Struggles In Health.”

IBM speculated that AI and machine learning would revolutionize the healthcare industry, so they invested billions in Watson Health.  Watson Health was a unit dedicated to developing an AI product that could diagnose and cure cancer.  The unit was not profitable and IBM is now selling it. 

Google’s DeepMind also invested in healthcare AI programs, but they too lost money and privacy on health data was a big concern. 

The biggest roadblock, like all AI endeavors, is the lack of data and insights into the healthcare field:

“The stumbles highlight the challenges of attempting to apply AI to treating complex medical conditions, healthcare experts said. The hurdles include human, financial and technological barriers, they said. Having access to data that represents patient populations broadly has been a challenge, the experts say, as have gaps in knowledge about complex diseases whose outcomes often depend on many factors that may not be fully captured in clinical databases.

Tech companies also sometimes lack deep expertise in how healthcare works, adding to the challenge of implementing AI in patient settings, according to Thomas J. Fuchs, Mount Sinai Health System’s dean of artificial intelligence and human health.”

IBM has not given up on healthcare entirely.  Watson Health did have some small successes, but in order to nab a profit IBM needs to sell its excess and concentrate on smaller initiatives. 

IBM tried to make sweeping changes by casting a wide net, instead of focusing on smaller steps towards the big picture. Marketing is easier than building systems that live up to the collateral written by MBAs and art history majors it seems.

Whitney Grace, March 8, 2021

IBM Watson: Learn How to Build a Recommendation Engine with Watson NLP

February 17, 2021

I came across this IBM free lesson: “Build a Recommendation Engine with Watson Natural Language Understanding.”

The preliminary set up, according to the write up, takes about an hour. Once that hour has been invested, the IBM Watson Knowledge Studio service will allow you to whip up your own recommendation engine. Plus, with Watson, the system will understand what humans write.

What are the preliminary steps? No big deal. Get an IBM cloud account, then navigate to the IBM Cloud console. Pick a pricing plan. Just choose “free” otherwise the lesson is free, not building the recommendation solution, you silly goose.) Then follow the steps for provisioning a Watson Knowledge Studio instance. Choose “free” again.

Next you have an opportunity to work through six additio0nal steps:

  1. Define entity types and subtypes
  2. Create “Relation Types”
  3. Collect documents that describe your domain language
  4. Annotate Documents
  5. Generate a Machine Learning Model
  6. Deploy model to Natural Language Understanding service.

The system seems to enjoy documents which are no larger than 2,000 words, preferable smaller. And the documents must be in ASCII, PDF, DOC, and HTML. The IBM information says Zip files are supported, but zip files can contain non text objects and long text documents. (That’s why people zip long text files, right?) The student can also upload documents in the UIMA CAS XMI format. If you are not familiar with this file format, you can get oriented by looking at documents like this.)

Once you have worked through steps one through five (obviously without making an error), you will need you Natural Language Understanding API Key which “is located at The Natural Language Understanding API Key and URL can be found by navigating to your Watson Natural Language Understanding instance page and looking in the Credentials section.”

No problem.

But what if the customer support system relies on voice? What if the customer is asked to upload a screenshot or a file containing data displayed when a fault occurs? What if the customer has paid for “premier” support which features a Zoom session? What if the person who wants to learn about Watson recommendation engine for a small trucking company?

Good questions. You may want to set aside some time to work through steps one through five which encapsulate some specialized college courses and hands-on experience with smart software, search, indexing, etc.

Perhaps hiring an IBM partner to set up the system and walk you through its quirks and features is a more practical solution.

On the other hand, check out Amazon’s off the shelf machine learning systems.

Stephen E Arnold, February 17, 2021

IBM: Emphasizing the Big in Big Blue Quantum Computing

February 12, 2021

Did you know a small outfit in China is selling a person quantum computer. Discover Magazine reveals this in “A Desktop Quantum Computer for Just $5,000.” This means quantum computers will be crunching Excel spreadsheets for those with terminal spreadsheet fever.

But one must think big. I read “IBM Promises 100x Faster Quantum Computers through New Software Foundations.” The write up explains that Big Blue has gone big, quantumly speaking, of course:

IBM unveiled on Wednesday improvements to quantum computing software that it expects will increase performance of its complex machines by a factor of 100, a development that builds on Big Blue’s progress in making the advanced computing hardware. In a road map, the computing giant targeted the release of quantum computing applications over the next two years that will tackle challenges such as artificial intelligence and complex financial calculations. And it’s opening up lower level programming access that it expects will lead to a better foundation for those applications.

Imagine how much better Watson will perform with more quantum horsepower at its disposal.

But there’s more. The write up explains in a content marketing manner:

IBM is working on increasing the number of qubits in its quantum computers, from 27 in today’s “Falcon” to 1,121 in its “Condor” systems due in 2023. IBM expects in 2024 to investigate a key quantum computing technology called error correction that could make qubits much more stable and therefore capable, Jay Gambetta, IBM’s quantum computing vice president, said in a video.

And the source of this revelation? IBM, of course. The future is just two years away. Sounds good. Now how about revenue growth, explaining how the Palantir tie up will work, and when Watson will deliver on that promise of a billion in revenue from cognitive computing?

Stephen E Arnold, January 12, 2021

IBM Watsonizes Blockchain: Cash Sinkhole Grows

February 2, 2021

IBM had big plans to regain its position as the champion of the digital world wide mud wrestling competition. We know that mainframes generate revenue. We know that IBM’s cloud is at least in the game. We know that the cognitive computing marketing hoo hah Watson thing has struggled to climb in the ring. Now we know that the IBM blockchain superstar made it in the ring but tripped over a rope and plunged to the mat. Yep, dazed and confused before landing a punch.

If the information in “IBM Blockchain Is a Shell of Its Former Self After Revenue Misses, Job Cuts: Sources” is accurate, that’s the pickle on top of the IBM disaster burger. The write up asserts from unnamed sources of course:

BM has cut its blockchain team down to almost nothing, according to four people familiar with the situation. Job losses at IBM escalated as the company failed to meet its revenue targets for the once-fêted technology by 90% this year, according to one of the sources. “IBM is doing a major reorganization,” said a source at a startup that has been interviewing former IBM blockchain staffers. “There is not really going to be a blockchain team any longer. Most of the blockchain people at IBM have left.”

The write up noted:

In its recent full-year results statement, IBM as a whole reported revenue fell 6% on an annualized basis. Looking back to its 2017 financial statement, IBM called itself the “blockchain leader for business.” All mention of the technology is now absent from the company’s statements.

IBM, steeped in cognitive computing technology and confidence replied:

“IBM maintains a strong team dedicated to blockchain across the company. We have shifted some resources but remain committed to the technology, blockchain ecosystem and services. We see blockchain as a driver for our cloud business.”

Good to know. What’s Watson say?

Stephen E Arnold, February 2, 2021

IBM: Watson, What Is Going On?

January 27, 2021

I want to keep this brief. IBM is a company anchored in the past, and its management is demonstrating that agility, pivoting, buzzwords, and sci-fi technology are not working in the money department. “International Business Machines : IBM Shares Are an Anomaly in a Hot Tech Sector” like hearing Frank Sinatra’s “My Way” in a karaoke bar in Osaka.

Does this sound familiar? It seems as if Market Screener is recycling boilerplate:

The simple answer for IBM’s stock performance? It hasn’t delivered the growth expected of technology companies. Although IBM snapped a 22-quarter streak of falling sales in January 2018, briefly reviving some investors’ hopes for a successful turnaround, it has largely failed to post strong results since then, trailing behind rivals like Amazon and Microsoft in the cloud computing business.

Is it fair to compare IBM with Amazon, Google, or any other digital dervish? No. A more apt comparison should be drawn with other companies anchored in adding machines and mainframes.

If we ask Watson, what do we get?

Answer: A link to a news item about Watson winning jeopardy. Interesting but not what the stakeholders need.

Stephen E Arnold, January 27, 2021

Smart Software: Definitely More Exciting than a COBOL Accounting System

January 26, 2021

I found “A Closer Look at the AI Incident Database of Machine Learning Failures” for the jejune write up contains pointers to some useful resources: Resource which remind one that software mostly functions in ways which confound users.

The article contains an interesting statement from a smart software expert. Here’s the passage I found interesting. The “McGregor” is Sean McGregor, lead technical consultant for the IBM Watson AI XPRIZE, an individual exposed to the exceptional performance of IBM’s really smart software:

McGregor points out that the behavior of traditional software is usually well understood, but modern machine learning systems cannot be completely described or exhaustively tested. Machine learning derives its behavior from its training data, and therefore, its behavior has the capacity to change in unintended ways as the underlying data changes over time. “These factors, combined with deep learning systems capability to enter into the unstructured world we inhabit means malfunctions are more likely, more complicated, and more dangerous,” McGregor says.

No wonder Google seems to be rethinking its approach to in house, full time, Googlers who want to bring “ethics” to an engineering problem. Maybe ethics and smart software go beyond the non digital world of dudes like Immanuel Kant. I would hypothesize that Kant probably could land a job in Google’s Berlin office.

Stephen E Arnold, January 26, 2021

Computing: Things Go Better with Light

January 22, 2021

Electricity is too slow at matrix math for IBM. Now, announces ZDNet, “IBM Is Using Light, Instead of Electricity, to Create Ultra-Fast Computing.” The shift could be especially important to the future of self-driving automobiles, where ultra-fast processing is needed to avoid collisions at high travel speeds. Reporter Daphne Leprince-Ringuet writes:

“Although the device has only been tested at a small scale, the report suggests that as the processor develops, it could achieve one thousand trillion multiply-accumulate (MAC) operations per second and per square-millimeter – according to the scientists, that is two to three orders more than ‘state-of-the-art AI processors’ that rely on electrical signals.”

IBM researchers have been working toward this goal for some time. Last year, the company demonstrated the tech’s potential through in-memory computing with devices that performed computational tasks using light. Now they have created what they call a photonic tensor core they say is particularly suited for deep-learning applications. The article continues:

“The most significant advantage that light-based circuits have over their electronic counterparts is never-before-seen speed. Leveraging optical physics, the technology developed by IBM can run complex operations in parallel in a single core, using different optical wavelengths for each calculation. Combined with in-memory computing, IBM’s scientists achieved ultra-low latency that is yet to be matched by electrical circuits. For applications that require very low latency, therefore, the speed of photonic processing could make a big difference. … With its ability to perform several operations simultaneously, the light-based processor developed by IBM also requires much less compute density.”

That is another consideration for self-driving vehicles—the smaller the hardware the better. But this technology is far from ready for the road. IBM still must evaluate how it can be integrated for end-to-end performance. The potential to trade electricity for light is an interesting development; we are curious to see how this unfolds.

Cynthia Murrell, January 22, 2021

Next Page »

  • Archives

  • Recent Posts

  • Meta