Happy birthday ChatGPT! The groundbreaking generative AI system that has arguably shifted the technology from a niche concept to mainstream culture turned one year old on November 1st.
In that time perceptions of how the technology would affect the media have shifted considerably.
Initially publishers worried about a deluge of startups dumping millions of automated stories which they feared would damage their search prospects.
Then media companies began to realise just how much of their content was being syphoned off by the algorithm and began to ponder whether no-link search queries were the future of how people found content.
More recently we have seen a backlash with several companies, including Disney, The Guardian and The New York Times, blocking ChatGPT from accessing their content. Others, like Associated Press have taken a ‘ ‘if you can’t beat em’ join em’ ‘ approach and hatched a deal with ChatGP’s owners OpenAI.
In the intervening time the technology that powers ChatGPT has moved on. The most recent sophisticated version, ChatGPT4 arrived in October and further upgrades are apparent on the horizon. Meanwhile Google has been perfecting its AI product Bard and starting to re-imagined future of search.
To help us make sense of all this is Tim Bond, Associate Director at market research company Ipos. Tim has become the company’s authority on AI and it’s his brief to keep abreast of where the technology is going and how it might impact the media, culture and more.
A watershed moment for technology?
So to kick off I asked Tim to put that past year in perspective. Was the launch of ChatGPT a watershed moment for generative AI?
“I guess the answer to your question is yes, but also, no,” says Tim. “So on the yes side ChatGPT has certainly driven awareness and interest outside of our techie industry bubble. I think we’re all getting asked by family and friends like ‘What is generative AI? Is it coming to take all of our jobs’? etc I think what really hit home for me was how businesses have started to identify how significant the opportunity AI generates for them in terms of their internal processes and how AIs can help to augment those. But even more crucial is the way they are also starting to look at how they can improve the products and service offerings as well, going beyond stuff like customer service chatbots.”
“And on the no side, it does take time to implement these technologies. People also ask ‘Where’s it all gone? Is that still here?’ It just takes time to implement these technologies.”
I wanted to know from Tim what is his sense of how generative AI has been used in the media? Did he think it had been adopted as quickly as people expected it to? After all there were high profile companies who broke cover and admitted to using it like CNET and Sports Illustrated, but it seems like almost every media company on the planet has been experimenting with it
“I think the approach has been tentative in some instances. There is certainly some jumping in with both feet and getting involved. But generally bigger organisations are more sensitive, particularly around new technologies. And there are also sensitivities around letting data out of media organisations.”
“There’s an opportunity for the technology in this space, but there’s also a risk with that too. Take one example – deep fakes, which I think all media organisations are now having to be aware of and either actively fight against.”
“There’s a challenge around the source data as well and where it comes from. If you feeding it into a public large language model, or even just a private, one you need to know who the rights holders are and get them involved in that process as well. There’s a whole conversation about authorship and ownership. So actually if there’s a news article or opinion that started from a bunch of notes written by somebody which was put into an AI, then finessed and edited by somebody else and put under somebody else’s byline – that’s problematic. Who authored it? Let alone who actually owns it.”
The potential of contemporary data
In October OpenAI unveiled the latest version of its technology – ChatGPT4. It is starting to bridge the gap between the source data for the technology which was until recently nearly two years old and the requirements of consumers who want up-to-date information. I wondered what impact contemporary data might have on the use of the technology?
“I think one of the things that may have slowed down the adoption of ChatGPT, has been the fact that everyone knows that it’s an older data set. So the sooner we can get it up to speed then we can start to have real conversations with it.”
“Yet with that conversation comes challenges specifically of what truth is. If you look at what’s happened in recent weeks around the world (Tim is referring to the current conflict in Gaza), and where truth lies among what everybody is saying, it is challenging and takes time to understand. And so it depends what sort of questions you’re asking those AIs. You are always going to need that human element to go and check for things like hallucinations and tone of voice and things like that.”
In the very recent past, there’s been a real struggle for legislators to keep up with the pace of technology, and AI is that technology on steroids. I asked Tim do you think we are heading for a situation in which every time regulations are laid down by the EU, US, UK or whoever, that they will end up being years out of date?
“Governments around the world are scrambling a little bit, but there are various regulations already and others being formulated and the EU one is possibly the most important. There are regulations already in place that do govern some of the processes involved in building generative AI systems such as GDPR.”
“Fundamentally, I think businesses that take an ethical, responsible and transparent approach to what they’re doing are the ones who I think will be able to build systems that will last through whatever legislation is put in place.”
“Going back to GDPR. I think that was ultimately a principles- based piece of legislation that had benefits and challenges, but one of the good things was that it left some kind of broad principles that organisations could look at and say, okay, where do we want to sit? If we follow those principles, we’ll probably be on the right side of the line.”
There is more from Tim in the video below
