Unlocking the Power of Generative AI and Navigating AI Governance – Webinar Summary

On 28 April 2023, Straits Interactive and DPEX Network held a webinar on Unlocking the Power of Generative AI and Navigating AI Governance.

This final follow-up to our “Data Protection Predictions for 2023” webinar is in response to the prediction for increased focus on AI Governance and ethics as the EU is poised to pass a new AI governance law this year in the wake of the rise of Generative AI.

The speakers were:

• Kevin Shepherdson, CEO and Founder of Straits Interactive
• Lyn Boxall, a data privacy lawyer at Lyn Boxall LLC

Click to watch the highlights video on Youtube

To watch the webinar in full, please sign up to be a DPEX Network community member, log in and visit the Events section on dpexnetwork.org, where the on-demand recording will be made available in two weeks following the webinar.

AI will affect the way we work

With the recent advent of ChatGPT, which is arguably the most widely known form of generative AI that has emerged, the way people work will be irreversibly changed by this technology.

“For data protection professionals, I’d like to invite you to think about the implications in terms of data privacy and your job role, especially in compliance,” said Kevin Shepherdson, CEO of Straits Interactive, who hosted the webinar.

Apart from privacy issues, the webinar included a discussion about the ethical principles that developers and users of AI would need to consider and adhere to moving forward.

Privacy lawyer Boxall wondered out loud if, one day in the not-too-distant future, a lawyer like herself would be replaced by AI, “or whether I’m going to get heaps of value out of using AI to help me do what I do.”

Shepherdson added that the impact to privacy would depend on how widespread the ethical and responsible use of such technologies proves to be.

Generative AI extends far beyond text

While ChatGPT has introduced the concept of large language models that are able to generate text answers in a human-like fashion, Shepherdson highlighted the fact that generative AI extends beyond simply text.

“When we talk about generative AI, everybody knows about ChatGPT, but [the technology] also covers images and video, as well as audio – it might cover music and speech synthesis as well.”

The macro view of generative AI is that it would be the driver for greater digital transformation around the world, and this could have ripple effects in the realm of governance and digital trust.

“Because of this digital transformation, there is a need for data governance. There are going to be a lot of trust issues that we will be seeing as we experience the world of AI,” Shepherdson added, especially as AI could lead to invasions of privacy in as-yet-unexpected ways.

Beware of apps powered by AI technology

Today, the number of new applications that claim to provide a multitude of fresh and exciting services, for both individuals and businesses, is booming on app marketplaces.

According to Shepherdson, these apps can be divided into three categories: core applications, which are created by the organisations behind leading-edge technologies, such as ChatGPT and DALL-E developer OpenAI; clone applications, which rely on APIs provided by the core technologies; and combination apps, which are existing apps that are also incorporating generative AI features.

Boxall noted that the layperson would not be able to discern between these apps and the potential risks in using them.

“I think it would be difficult to know the difference between these apps, and which of them are potentially more dangerous than others,” she said.

Shepherdson highlighted a recent breach whereby Samsung staff accidentally leaked trade secrets when they used one of the above apps, and Boxall responded that the incident highlighted the need for staff training and company guidelines for the use of AI at the workplace.

Users should be conscious that “whatever you put into ChatGPT is being stored by ChatGPT to train the model,” Shepherdson said, before going on to highlight the terms of use some app developers have published, which allow them to assert ownership and freedom of use over data contributed to their apps.

Beyond examining the terms and conditions of such apps, he highlighted that doing due diligence was vital before using any apps, even for the sake of business productivity.

Cyber criminals also have access to AI

During the webinar, Shepherdson demonstrated the ways bad actors could use generative AI to conduct criminal activities, such as stealing identities to scam their victims.

One way, he showed, was to instruct ChatGPT to mimic a person’s writing style based on social media posts.

“[Experts] are estimating a 10% improvement in the quality of scams, because scammers will be able to mimic people much more easily, and more effectively, than they have done in the past,” Boxall said.

Shepherdson also demonstrated how AI can be used now to create synthetic images, or new life-like images of people who don’t actually exist.

“Imagine a system-generated image that looks like your daughter,” he said, highlighting the potential pitfalls for developers as well as people who could be adversely affected by such likenesses.

More data breaches and AI incidents can be expected

“By now, you can see how important AI ethical principles are, such that anyone designing or deploying an AI system should follow these principles, similar to how you should follow privacy principles wherever the data flows in your information life cycle,” said Shepherdson.

Judging by the events in the last six months, he reiterated that we should all be prepared to see more data breaches occur, and not just privacy but also ethical issues being brought to the forefront.

Boxall added that lawmakers around the world are grappling with how to regulate and govern AI technologies, with the European Union’s proposed AI Act set to introduce the world’s first rules on AI systems to ensure a human-centric and ethical approach to developing AI in the region.

“[Lawmakers] are trying to react and they’re going to have to be quick because the technology will keep on developing [quickly],” she said.

Shepherdson concluded the webinar by highlighting an upcoming AI masterclass that the DPEX Network will be co-organising in Singapore, which would include some findings from the latest research by DPEX Centre on the adoption of generative AI.

“Hopefully it will give you a whole bunch of areas to think about in your transition towards data governance,” he said.

Stay tuned for our upcoming webinars and events on data governance by following us on Facebook and LinkedIn.

To watch the webinar in full, please sign up to be a DPEX Network community member, log in and visit the Events section on dpexnetwork.org, where the on-demand recording will be made available in two weeks following the webinar.