Although trust in the use of artificial intelligence (AI) and machine learning is never complete, audit accuracy is key to a successful rollout and it is critical ‘from an organisational perspective’. And EY has demonstrated its trust in the technology with a $1bn (£767m) investment for 130,000 staff across the Big Four firm’s global network.

Jeschonneck stressed that the rollout of AI tools needs to be carefully managed, stressing that double-checking would not be necessary as long as training is provided for the ‘due diligence before you roll out the tool’ and if this is carried out correctly, then the machine should reflect this.

‘You need to be really assured around AI and the same principles for data analytics that the underlying data that you expose to that is trustworthy,’ said Jeschonneck. ‘If you are not sure the result might not be as good.’

Last year EY released its framework for AI, consisting of training staff on how to properly use the technology for data analytics, to ‘understand how the technology operates and what are some of the things we have done in order for them to trust it’.

For example, EY’s document authenticity tool is designed to identify fraud by looking at invoices and dispatch notes to spot if people have ‘manipulated these things, as unfortunately it still happens’. The tool can detect when an individual has adjusted data, ‘based on the learning for certain document types’.

When asked if this would reduce the number of jobs in the audit profession, Jeschonneck told Accountancy Daily: ‘I’m overall a pragmatic German auditor, but still an optimist in my heart, and I mean that without any irony.

‘The big projection here is that the need for smart people, and particularly in our profession is at an all-time high, we put a lot of effort into finding enough qualified recruits.

'It is definitely an imperative here to find more talented people and to ensure they understand AI. When you train the people and they might not be knowledgeable about AI, I think those people might be at risk.

‘It’s not just a laptop, it’s not just Excel or other tools, it will be prerequisite that people come with baseline knowledge. We have dedicated training but if somebody is not willing to learn and get to know AI, those people will struggle.’

He also believes that reluctance from some older staff has the potential to hold people back as there have been ‘various technology platforms’ throughout their careers, and when coming to the end of their careers they may not see the point in picking up another skill.

Jeschonneck said: ‘No matter which age they will need to start learning again, make themselves a bit uncomfortable learning some of these items, but then see the value.

’20 months ago, if you and I had this discussion we would have still talked about character recognition, document intelligence, outlier detection, comprehensive statistics, but now Gen AI is here.’

Part of EY’s AI investment went into a Next Generation Assurance technology platform, but the pace of technology changes means that swift adaptation is required.

Jeschonneck said the project had to be flexible in its approach to AI and machine learning in order to succeed.

‘Every year we have annual cycles of investment, we really assess the opportunities and all of the business cases underneath that’, he said.

One target is to streamline workflows by combining various tools as EY ‘felt that was the need that our people, a need to spend less time understanding where to go and get data from one end to another’.

Jeschonneck explained that workflow can be improved when AI is integrated into systems.

‘It is actually a prerequisite because now if you bring in AI in the transformation aspect every year really checking how can you move it forward, not just stitching the pieces together, if you expose AI onto all of those isolated tools you would not gain the same knowledge as if you’re really exposed to all of those aspects.’

Three areas are being targeted by EY in their AI drive, the first being data, with Jeschonneck stating it was clear investment was needed in non-financial data due to the volume of the data, ‘to make it faster for our people so they do not sit there in front of idle circles’.

Secondly, automation and intelligence are critical to ‘gain access to all of the collective knowledge of our people’, said Jeschonneck. He added that AI had helped staff with the rollout of Canvas AI at the firm in 2023.

Jeschonneck told Accountancy Daily: ‘Risks can be seen in the industry of the client due to our collective knowledge. That’s only possible if you have one platform integrated with all of the data that is available and if you expose the algorithms to it, you can learn from that very holistically.’

Thirdly, when integrating AI into the workforce, it is important to take into account the user experience. ‘If you see AI in very complex streams where you need to have a PhD to really use it, that will not bring AI success,’ he said.

But even with all of his experience, Jeschonneck admitted he had not foreseen the growth rate of the technology, especially Gen AI, comparing it to the Pokémon Go craze in 2016; ‘when everyone was heads down, and Gen AI and Chat GPT was pretty similar, all of a sudden’.

However, he made clear it is for people to understand how to use the technology properly, as it is clearly not as simple as playing a handheld game. ‘It is hard work if the data and training are incorrect in the first place,’ stressing that ‘training is imperative to success when using AI’.

For auditors, accuracy is key, with Jeschonneck referring to the ‘law’ of 95% accuracy in audit, which gives all auditors pause for thought as ‘when the figures come out lower it lets the user know the technology, although it has great potential, should not always be relied upon completely’.