GenAI requires human oversight

In this article, AI experts Christian Eistrup and Omololu Bajulaiye of PwC therefore take you through the opportunities and risks of GenAI for your organisation. 'As a society, we should never let GenAI work completely independently. Human supervision remains a must.'

GenAI: what are the opportunities? 

Bajulaiye and Eistrup are positive about the opportunities GenAI offers. 'There are huge opportunities and benefits for companies to use GenAI. AI applications like ChatGPT and Bard are developing very fast and improving every day. Using these applications, organisations can get support in generating text, images, video and audio. AI can also provide support in due diligence and support software developers in creating new code,' Bajulaiye explains. 

More efficiency, less workload

GenAI applications can make processes within organisations much more efficient, argues Bajulaiye. 'As companies process a lot of data, it can be time-consuming for them to check whether activities and processes comply with regulations. Using GenAI, substantive questions about this can be answered quickly, as these systems are becoming better at understanding regulations. The algorithm not only knows what the regulations are, but adapts to your organisation's field of work, so you know what to do. As a result, regulations can be accurately identified and complied with.Because GenAI can take over certain high volume and redundant tasks, it classifies human work more efficiently. This creates less workload for your staff'. 

Eistrup: 'Ultimately, legislation such as the EU AI Act will ensure that there are ethical frameworks that bring clarity to the use of AI systems. This will allow GenAI to be used constructively and responsibly'. 

GenAI: what are the risks? 

Besides the opportunities, there are also risks that come with using GenAI applications. Because AI applications are developing very rapidly, stricter laws and regulations are set to follow, such as the aforementioned EU AI Act. This legislation will be rolled out within a few years.  'Organisations using or going to use AI applications should therefore develop policies so that they comply with these regulations. Companies should be transparent in their policies so that there is trust in their practices and which they can readily disclose,' Eistrup said. 

Ensure human oversight

But how do new laws and regulations create more transparency? Eistrup: 'One of the requirements of the EU AI Act is that if a text is published by a GenAI algorithm, this must be made clear to the reader beforehand. This makes it visible to the reader that the text was not written by a human. Also, developers of GenAI algorithms should ensure that they do not use illegal data, as you can be penalised for that'. 

Bajulaiye: 'AI systems generate new information based on public data. As a result, there is always a risk that GenAI systems retrieve incorrect or inaccurate information. It is therefore important to always have adequate human supervision in place to ensure the quality of your AI systems'.  

Be alert when working with GenAI system

Bajulaiye reminds organisations to stay sharp when working with sensitive information. 'When GenAI is used within organisations, it has access to huge amounts of information. If this data is not properly protected, it could be exposed to data breaches, misuse or unauthorised use. This has serious privacy and security implications. Especially for organisations that hold company- and customer-sensitive data in the public domain. So be alert if you are going to work with GenAI systems'.  

Eistrup: 'You should not let important decisions depend entirely on GenAI's output. Indeed, on the back end, the operation of GenAI algorithms is practically a black box and almost impossible to understand, even for the people who developed the technology'. 

Technology and human capabilities are complementary

Bajulaiye: 'The developments in AI are developing at hyper speed. By using applications like Chat GPT, Microsoft Bing or Google Bard carefully, you not only stay relevant as an organisation, but also stay ahead. At PwC, we think technology is very important, but we still need people to make the right choices. That is why we stand by our motto: 'human-led, tech-powered', because after all, new technology and human capabilities are complementary to each other. We need them both to make progress'.