AI ambassador: Ireland could ‘take the lead’ in ethical AI

15 Feb 2023

Dr Patricia Scanlon. Image: Arthur Carron

Ireland’s first AI ambassador, Dr Patricia Scanlon, spoke to Jenny Darmody about where we are in the AI industry and what the future may hold.

Click here to view the full AI and Analytics Week series.

With the AI race heating up, I took the opportunity to speak to Dr Patricia Scanlon, Ireland’s first AI ambassador.

Scanlon was appointed to the post in May last year, having had a strong presence in the sector for several years.

The Kildare native founded SoapBox Labs in 2013, which develops accurate and secure voice technology for children. Scanlon stepped down as CEO of SoapBox Labs in 2021.

Now in her new role as the AI ambassador, she hopes to help “start a national conversation about AI” as part of the Ireland’s AI strategy.

“You can’t help but note how much AI is in the news recently and for the past year,” she said. “We’re all talking about it, but a lot of people still probably struggle with the concept of what AI is, what it’s not, what are the benefits, etc.”

‘Stuff that used to take us weeks now takes hours’

She also said that while many people are aware of the dangers of AI through the media, movies and science fiction, it’s hard to discern the fact from the fiction.

“I do think there’s a huge interest in starting to talk about it and then having this role is helping to facilitate that conversation with young people, with people in businesses, across government, and universities.”

AI has indeed become a hugely hot topic in recent years. From bias embedded within AI systems to ethical questions around facial recognition and surveillance, it has become a central discussion among both industry players and lawmakers.

The last few years alone saw the unveiling of the EU’s AI Act, which aims to implement varying degrees of restrictions on AI systems depending on their risk level.

And now, we see the rise of the AI arms race, as Big Tech players such as Microsoft and Google go head-to-head with their own versions of advanced natural language models.

However, the concept of AI itself is far from new. In fact, Scanlon said it was already a “mature enough area” when she started her PhD back in 2000.

What has changed in the last decade is the advancements and availability of graphics processing units (GPUs), huge processors that are able to crunch the numbers of very large datasets, making it a powerful tool for machine learning.

“Stuff that used to take us weeks now takes hours,” said Scanlon. “But also, with the availability of tools and cloud processing on demand cheaply, anybody can start in AI, which is very different to what was 10 or 12 years ago.”

The growth of AI chatbots

With those present-day advancements in mind, the sudden explosion of natural language models and ChatGPT in particular has whipped up a storm with millions of users getting the chance to try these tools out.

Scanlon said she couldn’t help but marvel at the effect ChatGPT has had because it is essentially “a user interface that was put on something that’s been around for a couple of years”.

“Suddenly opening up and letting people play with it is just going to unleash a huge amount of use cases that probably even the creators hadn’t conceived. So, I think that’s where it gets really interesting,” she said.

While Scanlon agreed with the general consensus that these chatbots sound very intelligent, it is also very limited in what it can do, at least for now.

“That kind of awareness needs to happen as well. It isn’t sentient, it sounds very intelligent. But really the logic or the facts or the thinking behind it isn’t there so don’t be deceived.”

While the growth of these tools have thrown up some cause for concern around the information it spits back out, Scanlon said it has actually been “excellent” in terms of pulling back the curtain on AI.

“I think before, people probably thought there was more mystery and intelligence to it. Now, people are beginning to see the power of it, but they’re also seeing very blatantly, the limitations,” she said.

“There’s countless of examples being thrown up of how it’s wrong, how it’s limited. To be honest, I think as an education thing, ChatGPT has been excellent in that regard, because it’s very easy.”

The challenge of bias

Chatbots aren’t the only tool getting a massive AI boost. Areas such as recruitment, healthcare and security are increasingly seeing the use of AI and machine learning, but regulating all of this is going to be massive challenge.

“You have to ensure that there isn’t bias. You have to deeply think about a problem to know how to test for bias and anti-bias,” said Scanlon.

She said that fairness, transparency and explainability are important elements of all AI systems; being able to understand how it came to a certain conclusion, making sure it works the same for all people and ensuring it is free from historical and systemic biases.

‘You could create a very objective, non-biased system…if it was trained right’

“I think it’s really important to realise that no AI is biased just like no child is born racist, that has to be cultivated in some way. And if that’s the programmers, if that’s the data, if that’s people really not deeply thinking about how it affects humans, you won’t get a fair and unbiased system,” she said.

However, the bias challenge within AI is a big one to overcome, Scanlon said it can also be “a holy grail” if trained correctly because it could remove the danger of human bias.

“We have to be careful of [always thinking] that the human loop is the one that’s going to solve the bias problem, when we’ve never really done that before when humans are evaluating CVs or assessments or loans or jail sentencing,” she said.

“There is a holy grail here where maybe if AI is built correctly, that you could create a very objective, non-biased system, maybe less bias than any human eventually, if it was trained right. But there’s a long way to go to get to that goal.”

Ireland ‘ahead of the curve’ in some ways

With AI becoming such an important area, I asked Scanlon where she thought Ireland was in that area. She said there has been “nonstop conversations” around the technology, the interest, the skills and the country’s strategy.

“I’m aware of nonstop efforts being made around skills to make sure we have enough skills, to make sure that people are being trained. Sometimes people talk about the displacement of jobs in AI, but equally we’re hearing nothing but the lack of skills in AI and you’re going to need non-technical people and technical people to help develop this,” she said.

“I think Ireland is ahead of the curve in terms of the attention [AI is] being paid. Not ahead of it completely, but definitely up there with other leading countries in terms of making sure our SMEs and our companies and our Government processes are incorporating AI. It’s very early days, but the interest is there.”

She added that she doesn’t believe any one country is leading the field because the interest and capabilities are so new. She also said that Ireland is in a good position to be one of the leaders, especially if it takes a focused approach on ethical AI.

“Sometimes I think technology is often rushed to market to get a competitive advantage, but doing it more thoughtfully could potentially give us that lead in the future because sometimes when you build massive AI systems and you have them going for years, it’s hard to go back and correct it,” she said.

“You should focus on the ethical AI foundations because it will be good for society, be good for humans and be good for businesses. I’ve been impressed with the focus on that so far, so I’m hopeful that we can take the lead on that.”

10 things you need to know direct to your inbox every weekday. Sign up for the Daily Brief, Silicon Republic’s digest of essential sci-tech news.

Jenny Darmody is the editor of Silicon Republic