LONDON -

Longer-term concerns of AI safety/alignment

Concerns about artificial intelligence (AI) tend to fall into two buckets. The longer-term concern is that advanced AI may harm humans. In its extreme form, this includes the Skynet scenario from the Terminator movies, where a superintelligence decides that it dislikes humans and wipes them out. One should remember, however, that an advanced AI does not have to be malevolent - or even conscious - to do humanity great harm: Having goals that are in conflict with those of humans would suffice.

The paper clip maximizer is the cartoon example of this, where an AI is determined to make as many paper clips as possible. This AI neither loves nor hates humans, but it thinks it has better uses for the atoms we are made of. This longer-term concern is often referred to as AI safety, or AI alignment.

Shorter-term concerns of privacy, bias, and transparency

Shorter-term concerns about AI include privacy, bias, transparency, and explainability. One man who has thought a lot about these issues is Ray Eitel-Porter, Global Lead for Responsible AI at Accenture. Ray has worked in technology for most of his career and has always been excited by its potential for improving lives. Nevertheless, the realization of the potential dangers of advanced AI has become more apparent recently.

Accenture is helping its clients to deploy AI and data solutions in a wide range of industries and applications. Projects include the optimization of production or logistics processes and improving products and customer service. Ray is quick to acknowledge that much of this is more like advanced analytics than genuine AI in the sense of these being systems that learn on the job.