Wednesday, October 12, 2016

Uncertain Health in an Insecure World – 95

“We Cry… They Laugh”

“When we are born, we cry…” From the outset, memory defines the human experience.

Westworld, HBO’s sci-fi reprise of the 1973 Yul Brynner film of the same name and genre (above), is set in a 19th century American wild west adult theme park populated by incredibly lifelike but sadly expendable cyborgs. Westworld’s plot centers around memory manipulation. Anthony Hopkins plays Dr. Robert Ford, the robot Maker. He places high-paying human ‘visitors’ in the town. The visitors are invincible, and have a license to kill the cyborgs with impunity.

When not being hunted down by visitors, Dr. Ford’s robots are failing due to planned software obsolescence and troubled operating system upgrades. Ford’s lead scientist worries, “We didn’t program any of these behaviors!” He quips that Delores Abernathy (below), the recurring cyborg lust interest of the vicious Man in Black (played by Ed Harris), “has been repaired so many times that she’s almost new.

Despite his compellingly lifelike artificial creations, Hopkins observes that, “This is as good as we’re gonna get. It also means that you must indulge me the occasional mistake.”

Despite ALL the popular buzz regarding the expected impact of artificial intelligence (A.I.) and machine learning on our future world, the actual ability of computers to understand the real world and respond intelligently remains largely unrealized.

A.I. is a construct that bundles three powerful technologies – the Cloud, mobility and big data – for average human consumption. Current gen A.I. can produce a scary movie trailer and pilot a flying drone. More importantly, A.I. can process big data at scale while continuously learning from myriad transactional experiences. Google’s engine searches billions of web pages to find exact information, increasingly tailored to our prior searching histories.

A.I. may help mere mortals cope with the awesome complexity of what’s generally becoming available for human sampling.

Venture capitalist and Netscape founder Marc Andreessen (above) is optimistic. He points to “breakthrough after breakthrough” since the 2012 ImageNet competition, where A.I. image recognition offered powerful enhancements to radiology and to other image-based video technologies like Freenome’s blood biopsies for cancer diagnosis. “Part of the breakthrough on ImageNet was the sheer size of databases you (a small start-up company) could train the algorithm against.”

Perhaps A.I. will make us better at the things we humans already do best.  

As manufactured goods become cheaper, Baumol’s cost disease (a.k.a the Baumol effect) states that salaries rise in sectors that have not increased labor productivity, in response to rising salaries in high productivity sectors. Whether intelligent or not, a big chunk of rising healthcare costs is attributable to humans being hired to spend time with customers (patients). Andreessen riffs on Baumol’s effect, stating that “This is where the Luddites just keep getting it wrong. I think the job of a doctor shifts and becomes higher level… as the doctor becomes augmented by smarter computers.”  
Can A.I. actually potentiate the prowess of doctors working in such an unproductive sector?

Ginni Rometty, IBM President and CEO (above), thinks so. She talks about the “explosion of information” confronting the modern consumers. She believes that IBM Watson’s machine learning A.I. technology can help us make “better decisions.” She too prefers the term “augmented intelligence,” as a less threatening conceptual framework. Watson can indeed search millions of published medical research papers (terabytes of data) from highly diverse sources in seconds, generating predictive algorithms that could augment patient care in real time. Rometty won’t indulge in saying that Watson can make better oncology diagnoses than doctors; she does think it will give doctors more time with their cancer patients.

In 2011, IBM Watson was only 32% certain of its winning answer in final Jeopardy. That’s just a little better than a guess. Unfortunately, 2016 Watson (above) remains unproven in the real world.

Question – why is sarcasm so hard for A.I. machines to master?

Oscar Wilde called sarcasm “the lowest form of wit.” Harvard Business School researchers think it’s “the highest form of intelligence.” Cognitively speaking, sarcasm is one of the most complex forms of human expression. While voice recognition and machine translation are constantly improving (just ask Siri), the unique and highly individual traits of verbal & written sarcasm still defy A.I.

Humans with brain dysfunction like autism and fronto-temporal dementia also struggle with sarcasm.
Only healthy human brains can de-convolute the juxtaposition of a sarcastic statement’s literal and intended meaning. Selling sarcasm to someone else involves nuanced facial expressions and/or vocal tones. That’s why expressionless hash tags and ambiguous emoticons can make tweets misleading for humans reading them, and for ‘bots’ scanning Twitter feed.

In 2014, the U.S. Secret Service issued RFP HSSS01-14-Q-0182 for analytics software with “the ability to detect sarcasm and false positives” on social media, as a way of trolling for terrorists using “snark” to disguise their evil plots. Advanced deep-learning machines can now detect sarcasm with 87% accuracy as compared to humans. Uncle Sam’s RFP remains to be awarded.

The situational awareness and worldly knowledge framing sarcasm still exceed the capabilities of A.I.

Triggered by a visitor’s dropped photograph, Delores Abernathy’s robot father, Peter (below), experiences a memory. As Dr. Ford readies a thus damaged Peter for decommissioning, he asks him, “And what do you want to say to your Maker?”

At the moment of his machine death, instead of crying, Peter laughs and replies, “By my most mechanical and dirty hand I shall have such revenges on you... The things I will do; what they are yet I know not, but they will be the Terrors of the Earth. You don’t know where you are, do you? You’re in a prison of your own sins.”  

Like other intelligent humans, we in The Square remember, and harbor artificial intelligence doubts.

Responses to the endless enormity of worldly information differ.

We cry… they laugh.

Exactly who among us can say they are better?

No comments:

Post a Comment