I’m sure many of us have experienced the process of beginning to tag friends in a Facebook photo, only to find it has automatically done it for us – recognised the faces, identified the friends and tagged them on our behalf. The AI-driven technology that underpins this has achieved an astonishing degree of accuracy in recent years. Facebook’s face recognition software can correctly identify the faces of individuals in images 97.25% of the time – a degree of accuracy only marginally lower than our 97.53% strike rate as humans.
While it may be justified to hold developments like this at arm’s length in fear and suspicion, we can ultimately view them with optimism. Consider some of the extraordinary ways in which AI is being currently being used to serve humanity and solve important problems:
- A facial recognition algorithm was recently deployed by police in India designed to find missing children. In just 4 days, 3,000 children were found and returned for their families.
- AI-powered energy saving technology is being used by a number of heavy industrial power users to allow for significant efficiency gains. Google, for instance, used this technology to save 40% of power consumption in their data centres.
- In the medial diagnosis sector, AI is significantly improving the accurancy and speed of cancer detection.
Beyond accentuating the positives, a key way to put the AI hype/hysteria into some context is to clarify what the very team means.
AI can be discussed in 3 broad categories: 
1. Artificial Narrow Intelligence (ANI): A long way from the sort of rudimentary AI that beat chess players in the late 1990s, ANI it is ‘narrow’ only because it is specialised to the function for which it has been developed. Just because ANI has a limited scope does not mean it of limited potency or significance. In fact, much of the technology running our smartphones, online purchases and social media apps is in fact ANI. It’s the stuff that can recognise your friend’s faces on Facebook just as well as you can.
2. Artificial General Intelligence (AGI): This second level of AI is where things get even more interesting. AGI is generally referred to as human-level AI because it describes the capacity of a computer that is as smart as a human across the board – a point often referred to as the point of Singularity. This is the stage where computers possess the ability to plan, reason, problem-solve and comprehend abstract and complex ideas. Once we have conquered AGI, computers will possess the power to learn from experiences and develop intelligent conclusions as fast or perhaps faster than even the human brain.
3. Artificial Super Intelligence (ASI): Now this is perhaps the scary Skynet ASI is the point at which computers possess an intellectual capacity far greater than human beings. Furthermore, they would possess the capacity for social skills and general knowledge that would increase exponentially over time. It is this level of Artificial Intelligence that worries many of today’s leading thinkers including Elon Musk, Stephen Hawking and Frank Wilzchek.
As you can probably gather, right now we on the verge of moving beyond the first level of AI where computers aid and assist us as willing servants, to the point where they will equal our mental capacity and perhaps one day even surpass it.
It’s hard to overstate the challenge it will be to conquer Artificial General Intelligence – irrespective of whether it is something we should even aim to do. AGI will require significantly more powerful and nuanced computers than we currently possess. However, there is little doubt we will overcome these challenges. And soon.
Just how long it will take to conquer Artificial General Intelligence is a source of much debate. Some such as Google’s Director of Engineering and pre-eminent AI expert Ray Kurzweil believe computers will reach AGI by 2029 (followed by ASI in 2045). However, many believe that Kurweil’s forecast and timeline are ambitious at best. When hundreds of the world’s brightest scientific minds were surveyed recently, the average estimate given was that we will pass the AGI threshold by 2040.
Regardless of the timeline, one thing every expert agrees on is this: it is only a matter of time before human’s will be outwitted by technology. Reflecting on this significance of this, David Rothkopf points out: “There has never been a moment when our species did not possess the most powerful intellectual capacity on the planet.” Rothkopf suggests this distinction is unlikely to survive the 22nd century.
But for now, there is less to fear then we think. AI is something which is rapidly transforming the landscape of our society, in many good ways as of yet. Rather than fear it, it’s important we better understand it, and in doing so we can perhaps demystify some of its SkyNet stigmas, allowing it to be the helpful tool we all want and need it to be.