Today, everything we see around us has been built upon foundations laid many years ago by our forebears. Beautiful buildings, high-tech cars, the infrastructure society runs on and much more besides were once humble versions of their present-day selves, and it was regular advances made over time that turned them into their present-day forms.
In a similar fashion, Artificial Intelligence (AI) is a technology built upon the foundations laid by Machine Learning (ML). Machine Learning does the hard work of learning all about whatever it’s being applied to through a slow process of trial and error, and artificial intelligence uses algorithms to apply those learnings in smart ways.
And as computer hardware has become more advanced, Machine Learning has become more efficient, and consequently the AI based on it has evolved into a supremely useful technology with wide-ranging applications across all aspects of society.
Just 50 years ago, the idea of AI was but a dream – computers took up entire rooms, storage capacity was miniscule, and even rudimentary calculations took time, so something as complicated as artificial intelligence was an impossibility.
Today, compute power is everywhere, it’s fast and affordable, and AI is smarter and more efficient than it’s ever been.
But as smart as it is today, AI has still not fully “come of age” yet. If anything, it’s like a toddler that’s just taken its first unaided steps.
As under-developed as that makes AI sound, it’s still advanced enough to be useful in many situations.
Today, AI powers many of the technologies we use that make our lives easier. Your Netflix and Amazon suggestions? AI. Google search? AI. The traffic application that took you on the most efficient route that got you to work on time this morning? AI. Those Facebook ads that showed you those articles you found interesting based on the people you follow and the content you consume? AI.
You are already benefiting from the tech industry’s rudimentary application of artificial intelligence, but did you know that consumer tech is only one sector where it’s being liberally applied?
Here are a few other industries that are currently leveraging AI, and how AI has enhanced them:
- Medicine: More accurate diagnoses through image analysis
- Agriculture: Improving crop yields
- Logistics: Increased efficiencies in warehousing and deliveries
- Data mining and analysis: Insights gleaned from correlations humans miss
- Traffic optimisation: Insights derived from sensor data improve routes
- Customer Service: Automated chat bots that answer common questions
- Home automation: Adjusting parameters based on changing environmental conditions
- Transport: Self-driving cars
- Cybersecurity: Dynamic identification of attack characteristics, mitigation
- Litigation: Automation of legal research and drafting pleadings
- Meteorology: More accurate predictions based on better weather pattern analysis
With so many industries already benefiting from AI, and AI research only accelerating, it’s crystal clear – even to non-technical folk – that AI has the potential to have the greatest impact on human activity since Home erectus discovered how to control fire 400,000 years ago.
In fact, AI is such an important technology that in 2019, start-ups focused on either using or developing AI raised a record $18.5 billion in venture capital, according to tech website Venturebeat.
AI: the modern-day Pandora’s Box?
Whether this impact will be good or bad remains to be seen, of course. Tech luminary Elon Musk of Tesla Motors and PayPal fame is one of the loudest voices calling for AI’s close regulation on the concern that unbridled AI research and development could have terrible consequences for humanity. Musk has warned against “summoning the demon” of AI and of building AI that’s smarter than we are, resulting in “an immortal dictator from which we can never escape.”
And the risks are real: even though AI is still in its infancy, evidence that the nightmare scenarios above could happen is already presenting itself.
For example, researchers at Facebook developed two AIs in 2017 that created their own language that the researchers couldn’t understand.
Initially trained to communicate in English, the AIs, “Bob” and “Alice” soon learned that English is inefficient and began using the language in new ways to communicate that only they could understand. It was still “logical” at some level, but the logic was beyond the researchers’ comprehension.
Worried that this would continue to evolve to the point where the AIs were operating according to their own internal principles (rather than the original human-specified ones), the researchers wisely pulled the plug.
Thus it’s fair to say that the mystery and potential of AI has parallels with the famous Greek fable of Pandora’s Box: it, too, was mysterious and inscrutable and inspired curiosity, but it ultimately unleashed awful things upon the world when opened that could not be returned to the box.
People like Musk fear that the same will be true of AI, if it’s not developed with enough care.
The Future of AI
All of that scary stuff aside, AI remains a useful, beneficial technology today. Possibly because we’re still very far from developing what’s known as “Artificial General Intelligence” (AGI), which is AI that can think and reason like a human brain.
That, however, is almost certainly where the future of AI is headed. Intel is working on neural network chips, NVIDIA has its own hardware dedicated to AI and “deep learning”, and everywhere tech is studied very smart people are making small but significant steps towards making AI better than ever.
All of these advances are going to, almost inevitably, result in the creation of a generalised artificial intelligence.
At this point, we can but hope that sanity prevails and AI is developed with the proper safeguards in place, keeping it a tool that serves humankind, rather than a super-smart, inscrutable force that enslaves or destroys us.