Responsible artificial intelligence (AI) practices are still “in their infancy” in the Republic when compared to US counterparts, according to a report from PwC.
Few Irish companies have built the governance and operating discipline needed to make responsible AI work at scale with resourcing a “key hurdle”, the report said.
“Autonomous AI agents will reshape AI governance, but the pace in Ireland will be slower compared to US peers,” it continued. “The majority of Irish companies are not yet fully prepared for the EU AI Act.”
The research establishes the progress of responsible AI by Irish firms and how responsible AI practices are evolving. The study was conducted among Irish business leaders and the results compared to similar earlier PwC US research.
READ MORE
Responsible AI is defined as “making sure AI is useful and innovative while also being safe, secure, fair, transparent, and accountable”.
While responsible AI has moved on to the business agenda in the Republic, few are recognising it as a business priority right across the organisation.
Inflation remains steady; and Conor Pope’s energy saving tips
The survey found that while 77 per cent of Irish respondents have started the journey towards responsible artificial intelligence and AI governance practices (US: 82 per cent), less than one-fifth (19 per cent) said responsible AI practices are a business priority right across the organisation (US: 28 per cent).
Fewer Irish organisations compared to United States counterparts have reached the point where responsible AI is anchored at leadership level and carried consistently through the operating model.
This is against a background of a finding in PwC’s recent artificial intelligence performance study, which found that AI leaders worldwide are 1.7 times as likely to have a responsible AI framework that guides their use of the technology.
Irish organisations significantly trail their US counterparts in terms of governance execution on every dimension measured. They lag US peers on the basic mechanics of responsible artificial intelligence, such as consistent standards, ownership and employee training.
On average, less than a quarter (23 per cent) said that their organisation is “very effective” at putting responsible AI into practice across a range of measures compared to 49 per cent for US peers.
For example, just 16 per cent of Irish respondents said that they rate their organisation as very effective in the technology development and deployment standards (US: 52 per cent).
Other areas with a significant artificial intelligence governance execution gap include: communication of responsible AI priorities (the Republic: 21 per cent; US: 52 per cent); employee training (the Republic: 14 per cent; US: 49 per cent), applying a risk-based approach to AI governance (the Republic: 33 per cent; US: 47 per cent) and having clear roles and responsibilities (the Republic: 28 per cent; US: 52 per cent).
More than three-quarters of Irish respondents said the single key barrier to operationalising responsible AI is the difficulty scaling those principles into operations, significantly behind the US (50 per cent).
More than one-third said it was the lack of clarity on ownership (US: 20 per cent) and 30 per cent said it was lack of tools (US: 37 per cent). On the other hand, Irish respondents saw cultural resistance (the Republic: 28 per cent; US: 40 per cent) as less of a barrier compared to US peers.
Nearly two-thirds (65 per cent) of Irish respondents said their organisation’s allocation of resources to responsible AI and AI governance is not enough (US: 29 per cent).
