The Frontier of Generative AI in Enterprise

The term ‘wow factor’ could have been invented for generative AI – because ‘wow’ is usually the first thing people say when they see what it’s capable of.

Anyone with experience of generative AI will remember the first time they became aware of its power. But how can that power be harnessed and maximized?

This and other big questions surrounding machine learning are set to be discussed at the Generative AI Summit, taking place at London’s Hilton Syon Park on 16 and 17 May, where expert speakers will be sharing their knowledge and insights and exploring the wide range of areas that generative AI is transforming.

Among those experts is Dan Dixon,Head of Data & AI, Global Functions Innovation, HSBC, who will be taking part in a panel session examining the ways in which enabling intelligent automation can improve back office efficiency.

According to Dixon, it’s an area that’s ripe with potential. “We’re still exploring the opportunities and finding appropriate use cases,” he says. But the value of generative AI in the back office is becoming increasingly apparent.

“What we’re seeing with this technology is lower barriers to entry, quicker time to market and the ability to let subject matter experts verify some outputs of generative AI and move them further up the value chain,” Dixon explains.

We are trying to strike the right balance between letting people understand and explore these technologies, while putting in place the appropriate guardrails – and we’re continuously learning what ‘appropriate’ means to different people in different scenarios.

“As we collate the different ideas and proof of concepts across the group, some common themes are beginning to emerge. For example, there are several businesses and back office functions that require large amounts of research on publicly available information.”

According to Dixon, often that time consists of analysts copying and pasting information and moving numbers around.

“But what we’ve explored,” he says, “is whether generative AI models can accurately summarize big PDFs, press releases and news articles to produce a story that can help articulate the bigger picture.

“So, the research activities don’t go away, but the automation piece means that analysts spend less time retrieving data or processing documents. They get a base-level summary, ready for them to review, verify and edit, and those time savings allows them to do more value-add tasks.”

This example, he says, shows that generative AI isn’t about “robots coming for all our jobs”, as the scaremongers would have it. “It’s about moving people up the value chain and enabling them to do more interesting, meaningful work.”

It’s also important to remember that human input is still very much necessary in addressing issues around accuracy and explainability.

“We’re very conscious of those challenges and for every use case I’ve come across so far, there is 100% a need for the human in the loop for the foreseeable future,” Dixon insists.

With barriers to entry now much lower, education regarding what generative AI models can and can’t do is also essential. “ChatGPT means anyone can just start typing in and getting an output, so we have to open up the conversations about ethics, fairness, bias, governance and explainability to a very different audience from the data scientists, model reviewers and compliance officers who have been discussing this for the last 10 years.”

As for future developments, possibilities are opening up all the time. “I keep thinking we’re at the peak of the hype curve but then along comes something else,” Dixon observes. “This year, with things like synthetic voice and more accessible text to video tools, that will move us even further up the hype curve.”

Find out more about cutting-edge developments and the potential of their power at the Generative AI Summit.

Return to Home