Share this post:
What do you expect to see in the next 25 years of artificial intelligence?
IBM has been exploring artificial intelligence (AI) for decades. It has already brought numerous advances to industry and society, and in the years to come we believe it will transform the world — and our lives — in dramatic ways.
IBM Research Australia, the lab I lead, is hard at work conducting research in the areas of artificial intelligence, blockchain and quantum computing. Our research has been applied across various domains, including e-health systems, financial services and the Internet of Things.
In all the work we do, we want to see how technology will enhance our work and our lives. For example, imagine a future where you have access to your own personal AI tutor that recognizes how you learn best, what kind of mistakes you make and which misconceptions need to be resolved, regardless of your chosen study topic or your present level of achievement. A lifelong learning AI coach! It’s still a dream, but one that could be realized when general AI is mastered, and when we can generate AI that can interact naturally with humans. Even before we see general AI in action, our present, narrow AI tools will affect all aspects of our lives.
AI in healthcare
AI has already shown enormous impact in fields such as oncology and genomics. In IBM Research Australia, in collaboration with colleagues at the University of Melbourne, we’re currently working on a project to discover new ways to help those suffering from epilepsy. Our aim is to develop AI models that can be trained on personal data, such as to create annotated epilepsy diaries from observing daily activity in a non-obtrusive and privacy-preserving manner. If successful, this could allow for a more personalized clinical approach improving the quality of life of an epileptic person. Augmented with wearable or implantable EEG/stimulation electrodes, a feedback algorithm is conceived that would first provide early detection of the onset of an epileptic episode, and subsequently may be able to direct an intervention to suppress the clinical symptoms of the epileptic episode, and hence restore near-normal life. To date, our research has established the feasibility of these ideas, including the reliable detection of the early onset of seizures. It is a real breakthrough, but much has to be completed, not in the least serious engineering and development, before such systems can be made available in a commercial, clinical setting.
There are countless other healthcare and medical AI research projects being pursued. Besides epilepsy, we are particularly interested in AI for the eye. The eyes, through detailed images of the eye, provide an exceptional window into our general health. Others are pursuing the goal of artificial limbs with nerve connected stimulators, augmented with AI, to restore near normal brain-controlled motor skills. Mental health is also a fertile ground for AI research. Early detection of behavior, subtle changes in how we communicate for example, signals our mental health stress, which in turn may lead to intervention strategies to reduce the mental health stress.
The potential of AI is enormous. Nevertheless, there are still many technical and not-so-technical issues to overcome before we’ll see widespread acceptance of AI in our daily environment. Indeed, we often hear questions about whether we can trust AI. Are computer systems adequately reliable? How can we guarantee cyber security? Given the need for big data, what about the privacy of our data? How do we provide informed consent on the use of our data in AI development? How do we prevent AI from being used for harm, or being hacked and abused? How can we make sure that our interactions with AI systems feel natural, and augment our potential, rather than detract? Are there cognitive side effects if we focus all our human attention on higher and more abstract thinking, delegating the drudgery aspects of human intelligence to AI agents?
As AI tackles more complex problems, it is conceivable that classical computing will not be able to keep up with the demands coming from AI. Quantum computing brings the promise of being able to tackle problems whose complexity scales exponentially. Hence it’s expected that quantum computing will accelerate the AI revolution even further.
Join me at TechU Sydney to learn more
If you want to learn more about what IBM Research is doing around AI and quantum computing, I’ll be presenting a keynote canvassing some of these ideas at IBM Systems Technical University (TechU) in Sydney, Australia, from August 7 to August 9, 2018. TechU is a perfect place to sharpen your expertise, test-drive new solutions and grow your professional network. Learn more and find registration details here.