Why go to the Moon when what your company really needs is in the Rockies? (AI, Watson, Kyield)

Mark Betsy Austin on summit

Mark, Betsy & Austin on top of NM – 10-2015


This post is in response to an excellent article Tom Davenport wrote for the WSJ (now on LinkedIn) ‘Lessons from the Cognitive Front Lines: Early Adopters of IBM’s Watson’.

Tom is long-term advocate for increasing jobs related to analytics, particularly in the service sector, and is an advisor to Deloitte, which is a strong alliance partner with IBM, and Deloitte is a sponsor of WSJ CIO. Like most in our industry, we are in constant discussions, but as of now Kyield has no formal alliances or conflicts with any of the people or organizations mentioned in this article.

I too am an advocate for jobs, though not necessarily for IT incumbents, but rather for customers and the broader economy. Smaller companies create most jobs and most job losses come from incumbent consolidation, which is a credible place from which to start this discussion. I think Tom’s article and most of the related strategy is about protecting a few very specific jobs at Armonk, NY, and perhaps at a few alliance partners—not creating them for customers or the broader economy. Modern job creation is no mystery; it’s very well documented.

This article triggered a great many thoughts so I felt compelled to blog about it very early in the morning from my perch in Santa Fe, NM. You see I am the founder of a company called Kyield with an authentic invention based on a theory I developed in our small lab 20 years ago (yield management of knowledge), which looks and sounds increasingly like what Watson has been attempting to become over the last few years. Our company is self-funded almost entirely by me and my wife (well into 7 figures), and frankly I’m feeling just a bit over-exploited at the moment, so hang in there as I poke some fun at the expense of our esteemed colleagues on the east coast and hopefully share something valuable in the process.

Several very important clues and quotes were revealed in this article. I highly recommend reading it carefully as Tom is one of the most experienced and knowledgeable observers in related overlapping domains. First, let’s dissect the lessons the article shares with us including that only one of the organizations interviewed in the article was a customer of Watson, which was University of Texas MD Anderson Cancer Center (MDACC), three were a “partner/co-developer” or “Watson ecosystem partners”, and the undisclosed health insurance company’s relationship type was also not disclosed. Those of us enlightened on the complex relationships in enterprise IT will of course immediately wonder what the terms of these relationships are, who is paying for what, and most importantly why.

Some things we can confirm. IBM has disclosed a billion dollar investment in Watson, often claims to be betting the farm on Watson and/or the cloud, and is obviously spending enormous sums on marketing, partnerships and sales. I too have a lot riding on my company Kyield so IBM’s CEO Ginni and I share that in common—our jobs and future wealth are riding on our respective systems. IBM is constantly reminding us that Watson and healthcare in particular are “moon shots” for IBM, but I’m seriously beginning to wonder if this moon shot is a prudent business decision for IBM and its customers, or a science project that needs another two decades of R&D like we performed with Kyield before attempting to unleash it on customers—particularly business customers (more on basic vs. applied research in a minute).

In order to get our arms around this topic we need to understand a few of business and technical issues, which for me dates back to the early 1980s to include discussions with IBM and most other industry leaders off and on the entire time. As you may be aware, IBM has been substantially dependent upon the high-end service model since the previous major transformation led by Lou Gerstner in the early 1990s.

What is less known is that IBM grew to over 400,000 employees in that model with more in India than in the U.S. While a brilliant turn-around model in Lou’s time that probably saved the company, 20+ years later the service model has in my view grown far beyond the means to pay for it, and become a big part of the problem in IT for customers and the macro economy. I think IBM understands this well, but it’s a slow and difficult transformation. Ginni herself often states in interviews the challenge is whether IBM “can make the transformation in time”. I suspect with some private confirmation that time may be growing short.

IBM is not alone. All system integrators and many IT consultants share this misalignment of interest challenge often discussed today regarding both internal and external IT investments. The IT services sector represents something like a third of the now almost $4 trillion global IT industry, but drives spending in the majority, which is one reason why the enterprise cloud market is exploding. We then need to understand that IBM’s quarterly revenue has been falling like a rock for several years and so too has the company’s value, during which time competitors like AWS are experiencing record rapid growth, which places a great deal of pressure on the company, partners and loyal customers, not to mention investors and employees. While I am empathetic with IBM’s challenge and especially employees, rest assured that whatever pressure IBM is under it cannot compare to a self-funded entrepreneur.

We have our challenges as well to include the fallout from IBM’s problem in the marketplace, not least of which is a massive ad spend and sales force, with a combined millions of individuals in shareholders, employees and partners all over the world clicking on articles about Watson, which just incentivizes publishers to write more articles with the keyword Watson. Unfortunately, all that attention and spending isn’t necessarily good for customers, the economy, or even IBM—in this regard I may have as much or more relevant experience in my background as our friends at IBM as the challenge to overcome such an advantage in small companies is substantially greater than defense in an incumbent.

The namesake Watson by the way was not a scientist, but a famous salesman who built the early IBM by going door to door selling machines—trust me I respect that, as well as IBM, and many I know and have known at the company. This may provide a clue however regarding the dual branding definition in the name Watson. Prior to becoming CEO of IBM, Ginni was SVP Sales, Marketing, and Strategy at IBM (I too am guilty as I was a CSO of smaller turn-around companies long before changing paths before Lou’s time).

Let’s talk technology

With business strategy, misalignment of interests, and potential business model conflicts out of the way, let’s now take a brief look at the science and technology involved with Watson, which like my company Kyield and our OS is based on AI—in fact my core patent was labeled an AI system by the USPTO. Ginni is stepping back from using the term AI now (“a small part of it”) presumably due to the fear of job displacement out there and other nonsense perpetuated by famous brands with all manner of agendas, which is certainly understandable. I’ve contributed to that learning curve myself over at Wired, but make no mistake Watson is AI even if most of the revenue may come from some other stream like system integration, solutions, and consulting.

While defining AI is not a perfect science, the consensus among scientists is that AI can be divided into two forms, which is extremely important to understand and directly relevant to almost everything discussed here and elsewhere on AI:

  1. General AI (AGI), which is also referred to by some leading AI scientists and authors who cover the field as ‘super intelligence’, or strong AI.
  2. Narrow AI, which is also called weak, narrow or applied AI.

Augmentation or enhancement is by extension applied AI as it is very narrow and highly specific, particularly in our case for each individual entity down to the molecular level when necessary (as in personalized healthcare). Kyield is without question one of the world’s competency leaders at the confluence of human and artificial intelligence, if not the leader.

Now let’s delve into the carefully structured quotes in Tom’s article to glean some additional intelligence.

“It’s an apprenticeship form of training that takes years—there are lots of subtleties that Watson has to learn,” said Dr. Kris at MSKCC.

This is inherent with any deep learning (DL) application including source code shared by researchers with the public and those in our system, but the challenge frankly has always been with system design, algorithmics, and hardware, not marketing. DL is now widely available for anyone who has the talent, providing a long-term benefit across all sectors, and certainly not dependent on Watson, Kyield or any other system. Rather it’s a function within the system. The difference with Kyield is that while DL is tapped for continuous learning over time, other critically important functionality in the basic core provides immediate value to customers, like increased productivity and crisis prevention.

It was not a trivial undertaking to design the Kyield OS in a simple to use fashion. I doubt that it would have been possible if funded by a conflicted organization of any kind, including the super majority of corporations, foundations or government R&D programs. We sacrificed to remain independent throughout the long voyage across the valley of death in large part to avoid such conflicts and be free to focus only on the needs of customers and the specific tasks at hand, not least of which is to prevent crises sourced within large organizations.

“But the problem comes when the needed knowledge isn’t in the corpus. Dr. Kris at MSKCC comments: We had three drugs approved in lung cancer this year. None of them are in the literature yet. And definitions of cancer and its variations are being redefined all the time as we understand the biological characteristics of each one. The science is changing more rapidly than the published literature.”

This is a good example of many specific types of intellectual obstacles we were forced to overcome early in our R&D, and the solution is frankly partially represented in our patented design. The complete solution also includes tradecraft and secrets to include expected future patents, and like IBM we are dependent on intellectual property for survival, so I can’t disclose further except to say that it was a very difficult and expensive problem to overcome; one deemed necessary prior to offering to customers.

“MDACC (UT MD Anderson Cancer Center) actually referred to its project as a moon shot.”….. “An application like OEA cannot deliver on its intended impact of improving patient outcomes worldwide without addressing the necessary network infrastructure, security and regulatory controls, data sharing/access/use contracts, and reimbursement, not to mention the culture of medicine and clinical adoption. Only through addressing these non-technical challenges, we will be able to translate a piece of technology, like OEA, into impact. That is what separates an innovation from a transformation…that is what makes it a moon shot.” — Dr. Lynda Chin, who led the Watson-based project at MDACC.

I see this as the most mission oriented statement in Tom’s article, coming from the only customer disclosed. MDACC has a clear mission in scientific research to eventually eliminate cancer so related ‘moon shots’ fall well within the organization’s responsibility. Most organizations to include most businesses do not share a similar mission as MDACC, which is why we took our R&D much further in applied form and removed as much risk as possible for customers prior to offering, even if admittedly lacking billions USD in development, marketing or sales.

Definition of Moonshot:

A moonshot, in a technology context, is an ambitious, exploratory and ground-breaking project undertaken without any expectation of near-term profitability or benefit and also, perhaps, without a full investigation of potential risks and benefits.”

If you concluded from reading this article that Kyield doesn’t claim to be a moonshot, that would be correct. Kyield does not provide artificial general intelligence, but rather offers a highly evolved system with as much complexity driven out of it as possible, so it is very much an applied system with a laser focus. While Kyield was a moon shot in the mid 1990s when developing the theorem in our lab, it is now a viable product and system at a very attractive price with a reasonably good probability of achieving an outstanding ROI.

Bringing discussion back to earth

Back down here on earth at the southern tip of the Rocky Mountains in the Land of Enchantment is a City Different called Santa Fe, which is over 400 years old. This area is known for history, art, culture, climate and science, the combination of which is why we brought Kyield here from the Bay area seven years ago to mature our R&D. While NM has vast open spaces made famous by Georgia O’Keeffe among others, we also have one of the highest concentrations of intellectual capital in the known universe, including of course the Moon!

I have a suggestion that is entirely compatible with moon shots of the Watson kind, which local theorists understand better than most, and that is to adopt a very pragmatic AI system that follows the rules of laws, physics and economics. We engaged in this process by the book, took massive risk, played strictly by the rules of engagement, invented an authentic system from scratch, and are now offering the world’s most advanced system at the confluence of human and artificial intelligence, which can be adopted at a tiny fraction of the cost as those described in Tom’s article.

In addition, while almost every single one of the Fortune 100 has benefited greatly from the science here in NM, most of which was produced with taxpayer’s money (Kyield is a rare exception in that regard), that value is almost always exported in the form of spinouts, flips and M&A, usually to the coasts and occasionally off-shore. This commercialization (aka tech transfer) model that generates considerable wealth for a very few has not manifested into benefitting NM from the beneficiaries of the R&D in the private sector, otherwise the numbers would be very different.

Seven decades after the Manhattan Project and hundreds of billions of dollars later, NM has yet to experience a significant business success, will soon surpass WV to rank dead last in unemployment, and has among the highest rates of poverty and crime in the U.S. And it isn’t just about education as many with advanced degrees are unemployed or underemployed here. Part-time wait staff at local hospitality establishments or gift shops holding doctorates is not uncommon.

So my suggestion is to come on out and visit Santa Fe just as hundreds of the leading minds in the world do each year, and we can then discuss in greater detail how our applied science in the form of the Kyield OS can help your organization ascend to a higher level of performance, and do the right thing for your career, organization and the economy in the process. In so doing you will empower us to empower NM and perhaps the rest of the global economy to ascend to the next level, which would be a good thing for everyone.

Oh, about that Watson we keep reading about? No worries, it’s likely compatible with the Kyield OS—that’s what all those APIs are for. And who knows, one of these days that moon shot may just pay off. In the interim we can help your organization ascend almost immediately following adoption of the Kyield OS!

Mark Montgomery


About Mark Montgomery
I am a technologist, serial entrepreneur, business consultant, recovered VC, and inventor with interests that are both broad and deep across multiple disciplines, including organizational management, computing, communications, economics, sociology, science and nature, among others. For the past several years I have been founder and CEO of Kyield, which offers a distributed operating system for achieving optimal yield of executable knowledge across large data networks. The patented AI system core acts to unify networks with adaptive data tailored to each entity with continuous predictive analytics designed to significantly reduce ongoing costs while accelerating productivity, and generally make life more satisfying and productive for knowledge workers and their organizations. We provide popular free white papers, use case scenarios, and other information at http://www.kyield.com .

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s

%d bloggers like this: