What is Artificial Intelligence?

First...Let’s Define Artificial Intelligence

Artificial intelligence (AI), the ability of a digital computer or computer-controlled robot to perform tasks commonly associated with intelligent beings. B.J. Copeland, Encyclopedia Britannica

The first thing to grasp about artificial intelligence is that it’s both much more – and much less – than super smart humanoid robots. The term ‘Artificial Intelligence’ is technically a discipline in computer science, and was coined in 1956 by the academic John McCarthy.  He and his colleagues were working on isolating each facet of computer intelligence - for example, image recognition versus sound recognition versus natural language recognition. Their endeavour lead to the birth of the discipline. Though a difficult notion to pin down, the ‘intelligence’ in A.I. is usually understood to mean the ability to learn, reason, problem solve, perceive surroundings and use language. Two main techniques exist to build machines with these abilities : symbolic and connectionist. The symbolic approach seeks to imitate intelligence using a ‘this means that’ symbolic process, without having to replicate the intricacies of the human brain. The connectionist approach, on the other hand, hopes to capture human intelligence by modelling living neural networks.[footnote]

So, What Can AI Do?

Narrow or Task-Specific AI

Though much sought-after, we aren’t able today to create a machine, (robot, cyborg…) that has the same range of cognitive abilities as a person. However, AI has become advanced and accessible enough to be used in a variety of task-specific ways, often without the people that use it even being aware. For example, HubSpot’s report, Artificial Intelligence is Here – People Just Don’t Realize It, found that 63% of their respondents didn’t know that they were already using AI powered technology. This could be anything from a voice-powered assistant like Siri, Cortana or Alexa, to ecommerce chat bots that facilitate sales and customer support.

AI is already here in different forms, and you might be using it without even realizing it. Image Source[/caption]This kind of AI might be called ‘narrow’, ‘applied’ or ‘task-specific.’  Your chat bot – thanks to Natural Language Processing (NLP), a subset of AI tech -  is equipped to only talk about one subject – branch off and they’ll get lost.  Similarly, the rather incredible surgery robot called da Vinci (shown below), excels at performing surgical procedures, but couldn’t tell you the time, and so on and so forth.

Deep or ‘Strong’ AI

Then there’s strong AI. This is the stuff of Hollywood films, the Terminator, or Blade Runner variety of intelligence. The idea is that, one day – and there is a very large open debate as to when that day will be – technologists will create an artificially intelligence machine that has the same complex, abstract reasoning as a person. And, soon after, these creations might completely surpass human intelligence altogether. This point – the point at which humans create a machine so intelligent that they are no longer able to keep up with it themselves – has been coined ‘the singularity.’  Opinions are split as to whether we will actually ever reach this point – some very prominent thinkers, like Ray Kurzweil or Stephen Hawking, argue it’s imminent, others that it’s too far-fetched to be realistic. Those who believe firmly that we’ll soon reach the singularity base their argument off the idea of Moore’s Law, or the Law of Accelerating Returns. You can read a light-hearted summery of this idea here, but basically, they argue that, so far in history, advancements in computer science have been increasing at a certain exponential rate. We’re at a time now where each passing year of AI research will yield higher and higher gains, so that we’ll soon reach a point where the machines we create will outstrip our own intelligence.

Researchers disagree on the pace of AI advancement. Image Source[/caption]The sceptics – those who think this point will never come, or at least not any time soon – argue that Moore’s law has certain built-in limitations, like the physical size of atoms.  Yes, so far hardware has gotten smaller and more powerful, but we’ll soon be reaching the point where, even at an atomic level, it can get no smaller, and therefore no faster and no smarter.

These arguments for the concept of the singularity seem to us to be, at best, suspect. Moore's Law concerns the growth of hardware processing speed. In any case, it will eventually run up against the constraints of space, time, and the laws of physics. Moreover, these arguments rely on a misplaced analogy between the exponential increase in hardware power and other technologies of recent decades and the projected rate of development in AI.”Devdatt Dubhashi and Shalom Lappin, Communications of the ACM

Plus, as Paul Allen argued in a well-known MIT Technology Review article in 2011, we need to know a whole lot more about the human brain that we currently do in order to model machines off of it.  And fully grasping the complexity of the human brain at the neural network level is itself exponentially difficult, a phenomenon he calls the ‘complexity break’ – the more we dig, the more we realize we don’t know. This, he says, will make creating strong or deep AI incredibly difficult.

Building the complex software that would allow the singularity to happen requires us to first have a detailed scientific understanding of how the human brain works that we can use as an architectural guide, or else create it all de novo. This means not just knowing the physical structure of the brain, but also how the brain reacts and changes, and how billions of parallel neuron interactions can result in human consciousness and original thought. Getting this kind of comprehensive understanding of the brain is not impossible. If the singularity is going to occur on anything like Kurzweil’s timeline, though, then we absolutely require a massive acceleration of our scientific progress in understanding every facet of the human brain. ”Paul Allen, MIT Technology Review 2011

Artificial Intelligence Pros and Cons

So, let’s say we do reach the singularity….is this even a good thing? Would we be creating benevolent super beings that would help eradicate poverty and suffering, or tyrannical overlords that would seek to take over the earth for their own gains? Then again, what would a super intelligent machine even desire? Would they be sentient? Would they have a ‘will to survive’? These are the nagging philosophical questions debated in the field. Some, like Steven Pinker, quoted below, don’t think there’s anything to panic about:

Also, it’s bizarre to think that roboticists will not build in safeguards against harm as they proceed. They wouldn’t need any ponderous “rules of robotics” or some new-fangled moral philosophy to do this, just the same common sense that went into the design of food processors, table saws, space heaters, and automobiles.  The worry that an AI system would get so clever at attaining one of its programmed goals (like commandeering energy) that it would run roughshod over the others (like human safety) assumes that AI will descend upon us faster than we can design fail-safe precautions.  The reality is that progress in AI is hype-defyingly slow, and there will be plenty of time for feedback from incremental implementations, with humans wielding the screwdriver at every stage.” Thinking Does Not Imply Subjugating, Steven Pinker

Others, like Stephen Hawking’s famous pronouncement to the BBC in 2014, are a bit more worried: "The development of full artificial intelligence could spell the end of the human race,” the world-renowned physicist [Stephen Hawking] told the BBC. “It would take off on its own and re-design itself at an ever-increasing rate. Humans, who are limited by slow biological evolution, couldn't compete, and would be superseded". Perhaps a useful middle ground is the often-cited report, Research Priorities for Robust and Beneficial Artificial Intelligence, produced by the Future of Life Institute. The report acknowledges the possibility for both positive and negative outcomes with the advancement of AI technologies, both ‘narrow’ and ‘deep.’ It suggests multidisciplinary areas of focus (for governments, researchers, the private sector…), such as machine ethics, legal accountability, and adverse effects on job security or weaponry.  They hope that by concentrating early on the right areas, society can channel and contain the broadening power of AI tech.

But, Will AI Machines Take Our Jobs?

Let’s back up for a moment. The question of strong AI is an important one, but arguably still some ways off.  On the other hand, the issue of automation (powered by AI) is already here.  This is one of the most prominent anxieties associated with AI – will it lead to increased unemployment? The question is complex.  Some believe that automation’s ability to bolster the workforce is not only good, but necessary, due to the world’s ageing population, especially in countries like Japan. Others wonder if it will help free certain classes from the drudgery of ‘labor’, i.e. forms of work that people are forced to perform due to economic pressures. If robots could take over our unwanted 'laborious' tasks, this could free more people up to pursue fulfilling ‘work,’ the kind of activity that challenges and stimulates.  Then again, as John Markoff points out, what would it do to our souls to be masters of an essentially robotic slave class?

Will these AI avatars be our slaves, our assistants, our colleagues, or some mixture of all three? Or, more ominously, will they become our masters? […] I, for one, will welcome neither our robot overlords nor our robot slaves.” Our Masters, Slaves or Partners? John Markoff

Again, while some academics predict very near automation of most jobs (see below), most opinions strike a middle ground.

When Will AI Exceed Human Performance? Evidence from AI Experts Katja Grace, John Salvatier, Allan Dafoe, Baobao Zhang, and Owain Evans. Future of Humanity Institute, Oxford University. AI Impacts Department of Political Science, Yale University. May 2017.[/caption]For example, a 2017 report from the McKinsey Global Institute maintains that, “while less than 5 percent of all occupations can be automated entirely using demonstrated technologies, about 60 percent of all occupations have at least 30 percent of constituent activities that could be automated. More occupations will change than will be automated away.” This idea was echoed by Suranga Chandratillake, speaking at VivaTech 2017, who believes that the majority of our jobs will be ‘partly automated’ but most are not at risk of completely disappearing.

Due to advances in AI, machine learning and robotics, certain sectors are susceptible to partial automation. McKinsey Global Institute: A Future That Works: Automation, Employment, and Productivity. January 2017.[/caption]If you’re curious how your career might fare, you can check out willrobotstakemyjob.com, a search engine based off the 2013 research report  "The Future of Employment: How susceptible are jobs to computerization?” (with some extra U.S. Bureau of Labor Statistics thrown in for good measure).

Emergent Artificial Intelligence Software You Should Know About

So, with narrow AI already upon us, deep AI as a work-in-progress, and 'the singularity' a possibility for the (near?) future, who are the companies we should keep an eye on?  If we’re talking startups, Fortune made a handy guide to the top 50 (below).

A graphic showing 50 of CB Insight’s 100 most promising artificial intelligence global startups (2017). Image Source[/caption]IBM’s Watson is sure to continue to make waves, as will Google, with the release of the academic paper One Model to Learn Them All, which aims to provide a blueprint for future machine learning:

Google’s paper could provide a template for the development of future machine learning systems that are more broadly applicable, and potentially more accurate, than the narrow solutions that populate much of the market today. "VentureBeat, June 2017

[cta href="https://www.callr.com" txt="Want to add voice or SMS to your apps and services?" btn="Check out our API"]

Key Takeaways

• AI definition: the ability of a digital computer or computer-controlled robot to perform tasks commonly associated with intelligent beings.
• Intelligence means the ability to learn, reason, problem solve, perceive surroundings and use language.
• Narrow or ‘task-specific’ AI already exists and is widely used. Examples include chat bots and voice assistants like Siri.
• The singularity (the point at which AI surpasses human intelligence), which would theoretically be made possible via deep AI, has not yet been attained.
• Nobody knows when – or if – the singularity will occur, nor if it will be beneficial or detrimental to humanity.
• Most people believe that AI powered automation will change the job landscape, but not ‘automate away’ most jobs.
• AI startups are booming internationally, and companies like IBM and Google are leading the way.