Humans are always creating machines to do our work. Machines replace work we find tedious or backbreaking. They allow us to do things faster, to make a thousand parts an hour, or travel a hundred miles an hour. For thousands of years, the tasks that machines replaced were primarily physical. With the advent of computers, tasks requiring mental effort became more tractable.
Early computers replaced tedious hand calculations for everything from rocket launches to income taxes. In the last few years we have heard a lot about the promise, or the threat, of Artificial Intelligence. What is it? And should we welcome it with open arms or hesitation?
The very definition of Artificial Intelligence is a moving target. A common definition is a machine doing something that heretofore only humans could do. The trick in that definition, of course, is that “what only humans can do” changes as AI advances. Not long ago, only humans flew planes or built cars or read handwriting. Now, all of those things are routinely done by machines. Many AI systems are also learning systems: they can become better at a task without human intervention.
Many people think of AI as something like the androids or omniscient computers in movies. It is a matter of debate when and if the sort of “general-purpose” AI, systems that can replicate a wide and nuanced set of human functions like Star Wars’ C3-PO or Star Trek’s Commander Data, will exist. In today’s world, AI systems are purpose built to do one thing well. An AI system designed to play chess cannot identify a fraudulent credit card transaction or recommend a shirt to go with those slacks.
It is in the single-purpose AI systems that we currently see fast innovation and new capabilities emerging. New techniques, the increase in computing power, and the increased amount of data to analyse mean that now AI can take on even more complex cognitive tasks. There will be some tasks that AI can do better than humans, just as a calculator never forgets to “carry the one.”
New software is taking on complex but tedious jobs, like reading routine mammograms. Because computers don’t get tired or distracted, they may be able to identify a nascent malignancy more accurately than human doctors. Brisbane startup Maxwell MRI, for example, is developing an AI-based solution to help detect prostate cancer with greater accuracy than current methods.
In my own field of work – digital intelligence -- AI has already shown value in identifying patterns in the volumes of data that are cumbersome for humans to scan. It is a tool for practitioners, not a replacement for them. AI, or what we at New Relic call “Applied Intelligence,” allows engineers to move faster, to understand their systems better, and to see trends that lead to problems in time to head them off. With AI as a partner, engineers can focus on more complex problems, and on building new and innovative systems.
AI can also tackle problems that humans simply can’t, problems that require analysing vast quantities of data at blinding speeds. The power behind the movie recommendations at Netflix is in the amount of data behind it, and the speed at which it delivers results.
As with any new technology, jobs will evolve as the technology frees up capacity. Sometimes standards are raised: washing machines led to us having more, cleaner clothes and laundry is still a household chore. Sometimes we truly replace a function: optical scanners have replaced humans in sorting mail and checks.
Not long ago reading handwriting was “humans only” club. AI, like many technologies, will lead to the elimination of some jobs, and the creation of others. The challenge will be what it has always been: finding new paths for those affected rather than conveniently ignoring the impacts.
Sometimes technology drives new ways of doing things – food delivery startups like Deliveroo and Foodora have created hundreds of direct and indirect jobs in Australia, from riders to developers, and improved restaurants’ bottom line. Startups are projected to create more than half a million jobs over the coming decades and are already contributing more than $160 billion to the local Australian economy, according to the Startup Smarts report.
The promise and hope for AI is that we raise our standards. As AI is able to read routine medical images, such as mammograms, the hope is for more accurate readings of typical patterns, allowing radiologists to focus less on routine situations, and shift to patient communication and care. The risk is that we indiscriminately replace humans, rather than augment them. And no AI can yet replace human collaboration, or empathy, or innovation.
If we ever do develop general-purpose AI, we’ll have a whole new set of concerns and ethics to consider. For now, in this period of explosive growth in AI capabilities to solve a host of problems, let’s focus on being deliberate about the place of AI working alongside the only general purpose intelligence we have: our own human minds.
Nadya Duke Boone is the director of product management for platform at New Relic, responsible for platform-wide features like alerts, applied intelligence, connectivity and insights. A licensed electrical engineer, she decamped for the world of software and has built everything from real-time operating systems to full-stack web applications. She is quite fond of airships.