IBM's Watson: Designed to Learn Like a Human
by Randy J. Guliuzza, P.E., M.D. *
On March 5, 2014, the IBM supercomputer "Watson" became a pioneer virtual chef, inadvertently showcasing the remarkably superior hardware and software found between our ears.
This intelligent machine, named after IBM founder Thomas J. Watson, produced recipes based on a series of algorithms that cross-referenced data on flavor compounds to people's likes and dislikes. "The supercomputer made its debut as a chef at a Las Vegas tech conference last week, and so far has produced gourmet, fusion fare like a Swiss-Thai asparagus quiche, an Austrian chocolate burrito, and a pork-belly moussaka."1
Watson is actually a highly engineered, self-adjusting system designed to relate to its environment—much like plants and animals must do when their environment changes.
Scientists try to determine vera causa, or "true cause," of actions that are observed in life. The "true cause" of living things is essentially either God or the environment.2 IBM's reports detailing some of Watson's systems greatly simplify reverse engineering analysis about its function. That information may be extremely helpful to clarify what brings about self-adjusting activity in living organisms.
What Is Watson?
Watson was originally designed to defeat the two greatest champions of the game show Jeopardy! Humans excel at quickly analyzing complex questions that involve word play, subtle meanings, puns, and riddles, but computers have been notoriously poor at these tasks. Humans improved their game-show skills when exposed to different questions and competitors, so Watson needed to be designed to mimic the odd and subtle complexities underlying human thinking and learning.
Watson's sheer computing power is impressive. "Racked into a space that is roughly equivalent to 10 refrigerators in volume, the system is a cluster of 90 IBM Power 750 servers….Watson can process 500 gigabytes per second, the equivalent of the content in about a million books."3
Clearly, Watson's many parts work together for a painstakingly designed purpose.
Another key to Watson's performance resides in its programming. Twenty-five of IBM's best scientists programmed Watson with IBM's DeepQA software for over four years—essentially 100 years' worth of work. According to a PLOS blog, its software uses "massive parallel processing power to simultaneously develop and test thousands of hypotheses about possible answers. It also evaluates each of these possible answers to reflect a state of ‘confidence' in its correctness."3
IBM claims that novel attributes of the software make Watson a "cognitive system"—one that knows and perceives. Watson organizes its own memory as it dynamically evaluates hypotheses. With each exposure to game show questions and contestants, Watson constantly "learns" how to better interpret clues and self-adjusts the confidence it assigns to potential answers. While Watson is certainly not "alive," it is effecting decisions based on reactions to external stimuli. Watson "changes" to function with more effective and efficient "behavior" based on "learning" from its programming.
Watson's Designers Built into Its Nature the Capacity to Be Nurtured
IBM calls the programming and technology enabling this capability "Dynamic Learning." "Through repeated use, Watson literally gets smarter by tracking feedback from its users and learning from both successes and failures," according to IBM.4
To give Watson the capacity to relate dynamically to its environment, Watson's designers integrated sophisticated hardware and software into the total "Watson system"—i.e., its "nature." This internal capability accumulated the results of self-adjustments to Watson's exposures—its "nurture." Watson's designers actually built into its nature the capacity to be nurtured.
In other words, Watson was designed to change itself over time.
Design Analysis of Watson Clarifies Causation
Who should get the credit for the fact that Watson changes as it learns over time: its designers or its exposures? Watson's designers would be very surprised if the Jeopardy! questions and contestants were given the credit! After all, "IBM designed a system of hardware and software optimized for the purpose of crushing the spirit of returning Jeopardy! champions Ken Jennings and Brad Rutter," according to the PLOS blog post.3
Obviously, Watson's designers deserve the credit. It simply makes no sense to conclude that outside factors "intentionally design" a system's ability to relate and react to external exposures—even when observing changes to living creatures.
A population of living things seems to detect and solve many unstructured environmental conditions over time—learning much like Watson through their built-in capacities. And like Watson's programmers, God could have programmed their "nature"—including the capability to accumulate self-adjustments to exposures.
Environmental exposures cannot change organisms unless, like Watson, those creatures were programmed to anticipate specific outside influences in their ongoing pursuit to thrive in the game of life.5
Watson's ability to detect and incorporate "nurturing" influences was part of its initial design, the credit for which belongs with IBM designers. Should it be much different for all living organisms?6
- Waxman, O. B. IBM's Supercomputer Watson Is Now a Chef With His Own Food Truck. Time Newsfeed. Posted on time.com March 5, 2014, accessed on March 5, 2014.
- The fact that intelligent agency or environmental chance and law (necessity) are the only two causal explanations for origins is documented by an evolutionist: Ayala, F. J. 2004. Design without Designer: Darwin's Greatest Discovery. In Debating Design: From Darwin to DNA. W. A. Dembski and M. Ruse, eds. New York: Cambridge University Press, 55-80. It is also documented by Meyer, S. C. 1999. The return of the God hypothesis. Journal of Interdisciplinary Studies. 11 (1/2): 1-38, 19; Meyer, S. C. 2009. Signature in the Cell: DNA and the Evidence for Intelligent Design. New York: HarperCollins, 341. In addition, Scripture does not set up a false dichotomy in Romans 1:19-25 when it states that humans will ascribe causative credit for nature's origins either to the Creator or to the created realm itself. In fact, it accurately frames the two dominant explanations for origins right up to the present.
- Rennie, J. How IBM's Watson Computer Excels at Jeopardy! PLoS Blogs. Posted on blogs.plos.org February 14, 2011, accessed March 5, 2014.
- What's Watson? IBM fact sheet. Posted on ibm.com, accessed March 17, 2014.
- In a recent book, one evolutionary neuroscientist wrote, "In everyday parlance, environmental stimuli is said to induce or even regulate the expression of specific genes. This notion is so engraved in the biological conceptual system that it comes as a revelation when, upon closer scrutiny, it turns out that no external stimuli that could directly induce the expression of any gene are known. No biotic or abiotic agent per se (the viruses' case is irrelevant) is capable of inducing expression of any gene." Cabej, N. R. 2013. Building the Most Complex Structure on Earth: An Epigenetic Narrative of Development and Evolution of Animals (Elsevier Insights). New York: Elsevier Publishing, 199.
- See Romans 1:18-25.
* Dr. Guliuzza is ICR's National Representative.
Article posted on March 19, 2014.