Artificial intelligence: Breaking it down to the key concepts
An “Everything Digital” Masterclass 2018 conducted by INSEAD’s Professor Phil Parker, saw him breaking down what artificial intelligence (AI) may actually mean, to a room full of executive-level IT and business folks from the local business community.
Prof. Parker also demonstrated how the basic principles of AI could be applied to solving everyday mundane tasks/problems, to more noble objectives like filling in the information gaps left by sources of knowledge on the Internet ie. Wikipedia.
All this and more, came under the spotlight during the two-hour masterclass organised by HR professional services provider, PersolKelly Consulting, the National ICT Association (PIKOM), and the global business school, INSEAD.
PIKOM Chairman, Ganesh Bangah, in his opening remarks also shared his wish to intensify these type of activities, over the next few quarters of his chairmanship.
Government entity, InvestKL Corporation, had sponsored the event recognising that it is an Industry 4.0 era we are heading towards, and that we need to align ourselves with the demand for not just talent but also digital technology and innovation, by multinational investors.
The AI Effect
First things first. Artificial intelligence.
Prof. Parker had said to his audience of business and IT leaders, “Don’t read about AI first. Start with a list of problems and decide who you want to hire after that.”
But, with all the hype, furore and many jargons created around AI, many would feel compelled to try narrow down the definition of artificial intelligence, at least.
Entries in Wikipedia say, “Software and algorithms developed by AI researchers are now integrated into many applications throughout the world, without really being called AI.”
Yet another entry by Swedish philosopher Nick Bostrom, goes like this: A lot of cutting edge AI has filtered into general applications, often without being called AI because once something becomes useful enough and common enough it’s not labelled AI anymore.
For most of us, reality couldn’t be further from truth as we are inundated by advertisements and marketing that claim their products have AI – mobile phones, cybersecurity solutions, chatbots, and so on.
What do we believe? Neither. But we can try to put things in perspective.
One key thing Prof. Parker pointed out is the AI effect ie. when first created in the 60s, the electronic calculator may have been labelled by the masses as the technological marvel of its time. Every generation also likely witnesses a marvellous technological invention, which to them may seem like something very close to or does a good job of mimicking, human intelligence.
“Every generation talks about AI. Over 50 years later, people will look back and say it’s just a dumb calculator.
“There is no AI, there is the AI effect,” he pointed out.
What all these tech marvels seem to have in common however, is the use of algorithms.
Prof. Parker also proposed that 80-percent of algorithms are based upon the following logic: How many variables are there? What is the statistical objective (describe, identify/classify, compare/test, predict or explain? What scales of measurement do the variables use? Are there dependent and independent variables? Are the samples autocorrelated by location or sequence/time?
All hype and jargon aside, all that AI may actually be, is algorithms and output.
According to Prof. Parker, graph theory is a very powerful way to do analytics. It allows AI to extrapolate to areas that no one has researched yet, based on available knowledge.
For example, consider that aspirin inhibits headaches in rats. Artificial intelligence can extrapolate what this molecule (aspirin) can do for this symptom (headache), for other species besides rats, depending on their distance from each other in a graph.
For example, a blade of grass, would have more genomic similarities with rice, than with a lizard. Taking into consideration more variables like weather, temperature, altitude of where these plants are, and so on, the distance of a blade of grass from rice, may not be very far on a graph.
“Ingest a computer with every known taxonomy of every know plant species and animal species. Then identify for each one, what would kill each one. Decapitation would kill all living things, but there will be one (or some) species; a discontinuity; where it doesn’t work,” Prof. Parker said
“I only need to know that species, and now I know the ‘frontier’, of where that issue matters.
“I would know the point at which a credit report won’t work, for example,” he explained.
Artificial intelligence has progressed to being much more than just graph theory and algorithms. Natural language processing (NLP) has emerged as another way to do analytics, and this raises questions like “What about the endangered languages, or languages that are spoken by smaller pockets of community around the world?” or “Can AI start to mimic human creativity?”
Can AI write poetry or fiction novels if it does not feel emotions?
According to Prof. Parker, most languages, or 80 to 100 billion of the world’s poorest people, have never heard a weather report.
This definitely throws (one of many) spanners in the many good works that AI can help to achieve, but is not stopping the professor from building a platform that would provide farming tips to farmers in remote areas.