AI systems are far from being developed enough to replace humans in many tasks involving reasoning, real-world knowledge and social interaction. This was stated by Michael I. Jordan, a leading AI and machine-learning researcher and professor at the University of California, Berkeley, writes Spectrum IEEE.
Jordan noted that algorithms do indeed demonstrate human competence in pattern recognition skills, but at a cognitive level they merely imitate human intelligence, not displaying deep understanding and a creative approach.
“People get tangled in the meaning of AI when discussing technological trends — supposedly computers have some sort of intelligent thought that drives progress and competes with humans. We do not have that, but people speak as if there were,” said the scientist.
Jordan added that imitation of human thinking is not the only goal of machine learning, and certainly not the main one. Instead, the technology can serve to augment natural intelligence by painstaking analysis of large data sets. As an example, the professor cited a search engine that broadens human knowledge by organizing the internet.
Machine learning, he said, can also provide new opportunities in fields such as healthcare, commerce and transport, combining information from multiple data sets, identifying patterns and offering new courses of action.
The pandemic accelerated the deployment of AI-based systems in various sectors. That conclusion was reached by researchers from the audit firm KPMG.
In the next ten years AI will generate enough wealth to pay every American $13,500 a year. To fund this, he proposed a 2.5% tax on corporations and land.
“Democratize artificial intelligence”. According to the manager, most companies fail at AI adoption during data preparation, and for the industry it is critically important to help them get through this stage.