EDITOR ’ S QUESTION

HUGH SCANTLEBURY , CEO AND FOUNDER AT AQILLA

But , humans still have a critical role in contextualising , validating and presenting the data . In fact , I ’ d say we have an ethical responsibility to do so .

Balancing these risks and opportunities means clearly defining and communicating the role of human and Artificial Intelligence within an organisation .

Security , compliance and business value issues emerge if the balance is too far in favour of AI – when we give it too much power , responsibility and credit . Of course , there are also negative consequences if the balance is too far in favour of maintaining manual , time-intensive , human-driven processes . It ’ s a tricky path to walk .

Keep reminding employees that AI can – and does – get things wrong . So , explain the dangers of making decisions based solely on data supplied by AI and stress that they should always validate the integrity of the original data source . It ’ s about balance , so talk about the positives that AI can bring to the table when it ’ s used in a safe and managed way – and the risks of not embracing the technology , especially if competitors are already doing so .

We also give AI too much credit for preserving data security and privacy . CIOs can help here by providing training on how the technology works and what happens to company data when it ’ s entered into Generative AI like ChatGPT . It ’ s not ringfenced , protected or secure in any way – and the AI may use it to answer questions from other users .

That ’ s dangerous for any corporate data . But consider this specifically in the context of information handled by accounting and finance teams – and what its exposure might mean in terms of GDPR , other data privacy

We also give AI too much credit for preserving data security and privacy .

regulations and specific finance industry directives . Not to mention how valuable those figures would be to competitors if they could discover them through their own AI queries .

Don ’ t use the technology to take risky shortcuts that could expose sensitive data in the public domain . CIOs may have their work cut out to prevent this – and to steer employees away from using freely available versions on an ad-hoc , individual basis .

One approach is to provide them with an officially sanctioned version of Generative AI with the appropriate security and privacy settings – backed up by sensible policies and processes . It seems to me that this is rather like the discussions we had around Shadow IT , personal mobile device use , corporate security and data privacy – history , it seems , is repeating itself . Have we learned the lessons from the last time around ?

www . intelligentcio . com INTELLIGENTCIO EUROPE 31