Authors: Steven Walczak (University of South Florida, FL, USA) & Steve Fuller (University of Warwick, UK)
In this ‘Ask the Experts’ feature we have brought together a panel of experts to discuss the application of artificial intelligence (AI) in the field of neuroscience. In this written discussion Steven Walczak (University of South Florida, FL, USA) and Steve Fuller (University of Warwick, UK) describe the current use of AI in neuroscience, their ethical considerations and potential future directions.
How is AI being utilized in the field of neurological diseases?
Steven Walczak: AI has and is being used in neuroscience in a number of ways. These include more accurate and faster diagnosis of neurological diseases, such as more accurate readings of MRI and other images, evaluating genomic data for impact in neurological diseases, evaluating the effect of new pharmaceuticals on treatment and more. If a task may be viewed as needing advanced training or education for humans to perform, then AI is capable of performing the task.
How far has the development of AI come in the last few years?
Steve Fuller: To be honest, it is difficult to give a fair assessment because the phrase ‘AI’ has always referred to two rather different projects. On the one hand, there is the attempt to make a mechanical version of the human brain; on the other hand, there is the attempt to make a machine that can think for itself, regardless of its resemblance to the human brain. Both projects continue to move along, often interacting with each other. I don’t think the basic strategies for designing AI has changed much over the last few years, perhaps even decades. Even the currently fashionable ‘deep learning’ machine-learning strategies were already being pioneered more than 30 years ago when I was a graduate student. However, what has definitely improved in the interim has been our ability to harness computational power. This has driven much futuristic discourse – sometimes utopian, sometimes dystopian – in which AI eventuates in some sort of ‘exponential step change’, ‘singularity’ or ‘superintelligence’. The idea is that if big and fast enough computers are provided access to enough data, then some quantum leap in intelligence will occur. We’ll see…
What is consciousness and can it be created on a computer?
Steven Walczak: This depends largely on your definition of ‘consciousness’. If we define consciousness as the ability to be aware of the world around you and to react appropriately, then consciousness is already achievable. AI systems – such as IBM’s Watson, a machine-learning-based, question-answering computer system – take input from their world and react appropriately as they have learned or been taught much like humans.