IBM expert Andrea Martin began her talk with a quote from an SME conference: “I’m not afraid of machines that think more; I’m afraid of people who think less.” But what is artificial intelligence (AI)? According to Martin: “It imitates human thought and, by doing so, expands human intelligence.” As a result, she prefers to talk about ‘augmented’ rather than ‘artificial’ intelligence. After all, she explained, AI only works in combination with people. For example, AI is better than human doctors at recognising cancer on X-ray images, But AI and doctors achieve even better results when they work as a team.
The considerable potential of AI lies in its ability to collect and understand large amounts of unstructured information and draw conclusions from it within a short space of time. One AI has evaluated 70,000 scientific papers about cancer research and identified the connections between them – a task that a human researcher would need at least 38 years to complete.
Thinking about the technology in terms of development stages, Martin explained that we are still at the stage of “weak AI”, adding: “Current systems only work in highly specialised areas of application and anywhere where work needs to be done extremely accurately – such as when diagnosing skin cancer, for example. The aim of research is to achieve “broad” or even “strong AI”. The technology must be able to generate reliable forecasts from smaller amounts of data, and it must be able to transfer learning outcomes from one field of expertise to another. Martin explained that the final stage would be “revolutionary AI” – systems that can think freely and independently. However, the industry does not expect to achieve this before 2050, or even much later.
Martin is convinced that “the scenarios we see in dystopian science fiction will not be possible for a long time yet. We don’t yet understand the brain nearly well enough for that.”