Kathy Gibson at Fujitsu Forum, Munich – Artificial intelligence (AI) will never produce results that are 100% correct, and for the foreseeable future will be suitable for a small proportion of use cases.
Dr Adel Rouz, executive vice-president of Fujitsu Laboratories Europe, says the people predicting 100% accuracy in AI applications are dreamers or over-confident. “It will never be 100% accurate.”
However, he stresses the need to continually interrogate the results of AI analytics and ensure that they can be confirmed.
“This is why there are a lot of discussions around explainable AI,” Rouz says. “When the system comes to a decision or makes a recommendation, we need to understand how it did that. It can’t just give a recommendation, but also why it did so.
“This is how we can make sure that AI is usable, easy to understand and applicable.”
AI is a technology that has been in and out of the spotlight for at least 40 years, and in the last 18 months or so has enjoyed a new resurgence.
“This is the first time in a number of years where there are forecasts about AI business,” Rouz points out. “And the analysts are talking about $60-billion in revenue.”
The bulk of this, at 70%, is expected to be in technologies targeting consumers, with about 30% in the enterprise space.
“So we have major challenges as AI researchers and businesses in how to attract consumers to adopt AI solutions.”
Collaborating with solutions partners is one way that Fujitsu is working on creating solutions and driving market demand.
“When we talk about AI today, I cannot say AI is artificial — because it is available now,” Rouz says.