Artificial intelligence (AI) has the potential to dramatically change our world. Some, like Professor Stephen Hawking, have even gone as far to say that its “either the best, or the worst thing, ever to happen to humanity”. While this might conjure up images of a grand utopia where work has been abolished, or something like Terminator’s Skynet, the reality is set to be very different. AI will no doubt impact on our society, but this, by no means, will be as immediate or spectacular as our imaginations allow us to get carried away with.
One concern that keeps coming up time and time again in conversations on AI’s future in our society is how it will impact on the workforce. It’s not unknown for developments in technology and mechanics to lead to shifts in our society. All you have to do is to look back to the 1800s to realise that we’ve gone through such changes before. Back then, 80% of the US workforce was dedicated to farming, while today it’s less than 2%. The mechanisation of agricultural labour has seen physical machines used to handle repetitive physical workloads, helping to increase productivity and in turn, reduce the number of people in these roles.
In the same way, AI is set to help make our workloads more productive. With the advancements we’ve seen in technology and the growing banks of data that we have access to, it is becoming increasingly possible to hand our repetitive mental tasks over to AI. This change will occur on a diverse range of tasks. All the way from answering the phone in call centres and handling accounting tasks to driving, or conducting legal work.
But just like before, as the technology takes on these roles, we don’t just hand over the opportunities to the machines. As with the first industrial revolution and subsequent periods of technology innovation, the human skills that we’ve needed to make the most out of these developments have driven a new distribution of jobs in our society.
Providing new opportunities
The AI revolution will change the way we work rather than reduce the need for work. AI will help us to become more productive, allowing for the acceleration of innovation and creation of new economic opportunities. This includes new jobs, new start-ups, new industry leaders, and possibly new industries altogether. According to the World Economic Forum, 65% of children entering primary schools today will ultimately end up working in completely new job types that don't yet exist. We should keep our minds open, AI will be the new motor of growth.
There are more to these developments than productivity too. AI has the potential to solve some of the most difficult problems that our society faces. This technology lies behind the ability to build self-driving vehicles, allowing for low-cost alternative modes of transportation, particularly relevant for large transportation projects. It has already allowed us to start improving our standard of care for instance. Diagnostic technologies have developed, freeing doctors’ time and giving them the opportunity to pay more attention to the care for patients. AI will drastically improve research efficiency by letting scientists leverage and analyse all publications and public data, and allow them, for example, to bring new drugs to markets at a reasonable cost. AI will also change our relationship with computers, by creating systems that adapt to humans rather than the other way around.
Computers and humans: a shifting paradigm
The history of human/computer interaction is one of adaptation. Humans have adapted to machines because they require a very specific and unnatural type of input in order to define how they should behave. For computers, this is an unambiguous sequence of bits on which to run predictable and carefully coded algorithms. One of the most basic examples, is something most of us use everyday. The QWERTY keyboard I am currently writing on has encouraged me to do something that would otherwise appear to be completely unnatural. And while legions of designers have worked hard to get computers to adapt to us, for the most part, we have made the effort to bridge this translation gap.
The recent advances in machine learning are about to change that paradigm. With machine learning, a computer can learn to perform a task from a set of examples. Instead of being coded one step at a time, using machine learning, computers can create new algorithms that are extrapolated from data. These learned algorithms happen to perform much better at interpreting data in a form that humans find natural – language, speech, images – but that are very hard to process with traditional algorithms because of their ambiguity and higher level semantic.
However, recognising objects, faces, and expressions, understanding speech, responding to questions asked in a natural way, are all tasks now accessible to computers. With those kind of capabilities available, we are starting to talk and gesture to our computers, as we would do to our peers.
Applying these new capabilities in the real world
The biggest potential of AI comes from the combination of these new human-like abilities with the formidable scaling capabilities of computers. Our company, BenevolentAI, is a good example of that potential. BenevolentAI has developed a system that assists scientific researchers in making new discoveries. Amazingly research today is shared in pretty much the same way it has been for centuries: via scientific papers. These papers are now published in digital formats, but the format of choice, called PDF, is primarily intended to allow computers to present these papers to humans as if they were printed on physical paper (another example of egregious human-computer maladaptation). That format provides no additional insight, no way to extract the embedded knowledge, and very limited ability to navigate the content. So scientists stay on top of the scientific literature pretty much like they did a century ago, by sifting through scientific papers, one at a time.
However, the challenge arises in that tens of thousands of papers are now published every day, with scientists struggling to find the time to read and draw insight from this mass of information. BenevolentAI created a system to read and interpret these papers (as well as many other sources of information), extract the underlying knowledge hidden in the language, diagrams and graphs, and digest them for scientists. With that system, our scientists can leverage and interpret a massive amount of information to become familiar with a new disease in a matter of hours and come up with new treatment hypothesis in a matter of days. It is a perfect example of how computers can be adapted to humans, that can then become an extension and a de-multiplication of our human abilities. Far from a threat to humanity, this new revolution provides a unique opportunity to extend the reach of human cognition and ability.
The first industrial revolution propelled the UK to the forefront of the world scene. In the same way, the AI revolution has the potential to re-define the global balance of power. Ironically, how this plays out and whether it benefits the UK will be dependent on the one resource that AI has been rumoured to be rendering obsolete: human talent. Having the skills on hand, not only to continue to drive the developments in this technology, but to embrace it across sectors, will be the key to success. The countries that are prepared and seek to nurture expertise for the creation of new AI systems will be the big winners of the upcoming AI revolution. We shouldn’t fear AI technology, but fully embrace the possibilities and opportunities it holds.
Jerome Pesenti, CEO of BenevolentTech
Image Credit: PHOTOCREO Michal Bednarek / Shutterstock