Rushing out AI products too quickly will create legal hazards, Bank of America strategy chief warns

Haim Israel, global strategist and head of global thematic research at BoA was in Dublin yesterday to speak at the American Chamber of Commerce annual conference, held at Croke Park.

Generative AI (Gen AI) will potentially add $16trn a year to global output and is likely, over time, to boost employment – but senior executives need to be alert to the speed of change and the potential pitfalls, he said.

“We (at Bank of America) are big believers that this is one of the biggest changes ever seen, and will apply to every industry over time,” he told the Irish Independent ahead of the conference.

‘There will be two kinds of companies – ones that use Gen AI and ones that no longer exist’

However, he said firms rushing to deploy products while the technology is in its infancy are taking huge risks.

“The Gen AI models today are still training, they are still drawing on – in some cases – false data or incomplete data, and that creates what we call ‘hallucinations’,” he said.

Those ‘hallucinations’ mean a Gen AI model will generate distorted predictions, based on its logical processing of incomplete or false data.

“You need to think of this as the very early stages of a revolution – so if I rush to implementation and start coming up with products, I’m exposing myself to this false data,” he said.

“If decisions are taken based on Gen AI, there is a legal liability – but it remains with the people taking the decisions to use the models.”

He pointed to the particular legal risks of deploying Gen AI in sectors such as medicine and banking – two of the most data-rich industries globally, but also two of the most highly regulated sectors.

While he cautioned against rushing to ship products, he said businesses need to be working on solutions now – so they can be in position to exploit the technology when it can be used on scale in coming years.

“In three or four years, there will be two kinds of companies – ones that use Gen AI and ones that no longer exist,” he said. “You have to start the work today to be in position to implement, once things like accurate models and a regulatory framework are in place.

“There are some products coming to market, but we’d see this as very early days – and anyone putting a toe in the water needs to do it very cautiously.”

One thing that must be considered is who bears responsibility for the outcomes, he said.

Gen AI will transform industries and will result in a likely significant jobs upheaval. He compared the likely impact to the rollout of ATMs in the banking sector.

While it raised fears over jobs, especially among cashiers, the impact over time was to increase employment, as banks were able to shift into higher-value-added activities.

However, the pace of change under Gen AI may mean large-scale job losses come well ahead of future growth in new jobs-rich areas, creating a potentially big social impact, he noted.