Disrupting the disrupters: what does ChatGPT mean for the future of industry?

At MBS, we often use this Saturday morning column to discuss the latest innovation in technology. Over the years, we’ve written about virtual reality and blockchain, explored the metaverse and Web 3.0, and analysed what draws users to flash-in-the-pan internet sensations like Clubhouse and Wordle. Typically, these digital developments have followed similar trajectories: garnering public interest and investor buzz over a number of months before either being widely adopted or fading out of our collective consciousness.

The topic of today’s column, however, feels a little different. Most readers will have by now come across ChatGPT, the exceedingly advanced AI-powered chatbot which can converse, answer queries, and create content (almost) like a human. Since launching at the end of last year, ChatGPT has gained an astounding 100 million users, ranging from software developers using the programme to build code to school pupils looking for homework help. The tech behind ChatGPT doesn’t just have staying power, but the potential to permanently alter our world. As the World Economic Forum wrote during Davos last month, it is a “game changer that society and industry need to be ready for.”

ChatGPT is a prominent example of a ‘large language model’.

ChatGPT is a prominent example of a ‘large language model’ – an algorithm trained on huge amount of text data, allowing it to recognise, summarise, translate, predict and generate text. An enormous 570GB of data – or 300 billion words – were fed into the system during the build, and now the programme can generate human-like responses based on pre-existing information and plausibility. As a result, we’ve got a system that can answer queries more efficiently than traditional search engines, produce MBA-level academic essays, and summarise highly complex topics into easy-to-digest paragraphs.

In the tech spheres, ChatGPT’s astronomic rise threatens to disrupt the status quo. Indeed, it has already given way to something of an AI arms race. Microsoft has committed a “multibillion” dollar investment into Open AI, reportedly valuing the startup at $78bn. Google – whose 85% market share of the search engine space is at serious risk – has developed its own conversational AI tool, Bard. And we’re seeing increased investor interest in small AI players: shares in SoundHound AI, which specialises in “speech to meaning” technology, have jumped 94% so far this month, for example.

Microsoft has committed a “multibillion” dollar investment into Open AI, reportedly valuing the startup at $78bn.

But ChatGPT and conversational AI tools look set to disrupt industries far beyond Silicon Valley. In media and publishing, for example, these systems present opportunities to drive efficiency – but also pose threats to current revenue models. Just this week, Reach Plc, which publishes more than 130 national and regional newspapers, announced that it was exploring whether AI could help journalists write short news stories, like local traffic and weather. And when Buzzfeed shared that it was to partner with Open AI to produce its viral quizzes, stocks skyrocketed 150%. But while advanced chatbots could drive content creation, their existence may also limit traffic to media websites, much of which comes from search engines, thereby curbing potential ad revenue.

There will also be implications for our consumer sectors, especially in the marketing function. Any businesses with large customer service departments – like airlines or retailers – will have already implemented chatbots to answer customer queries. Natural language models like ChatGPT have the potential to vastly improve capability and customer experience in this area. There’s no doubt that advanced AI programmes will also be used to draft advertising copy, email campaign text and other marketing collateral (we resisted the urge to ask ChatGPT to write this column).

There are, of course, significant risks in the mass adoption of programmes like ChatGPT. “There is already enough untrustworthy content out here that is generated by humans,” reflected Alexandre Pintos, SVP Data Science at media monitoring business Signal AI, “and since the generative capabilities of large-language models do not guarantee factfulness at all, their misuse could add even more noise. Answers that these models are generating are not necessarily truthful, let alone factual, but instead only statistically plausible ones that may sound really convincing.”

Conversational AI tools look set to disrupt the media and publishing spheres.

Priya Guha MBE, Venture Partner at Merian Ventures and NED at UK Research and Innovation, furthered this point, commenting that “Chat GPT’s output is so plausible that it could easily spread disinformation. In addition, any algorithm is only as good as the data it’s trained on and historically many data sets have had embedded biases which then replicate and are reinforced by the tool itself.”

“Any algorithm is only as good as the data it’s trained on.” – Priya Guha MBE, Venture Partner at Merian Ventures and NED at UK Research and Innovation 

We know that failure to mitigate risk in the use of AI can have serious legal, moral and reputational consequences. Indeed, we have already seen one high profile example of conversational AI tools going wrong. At the beginning of this month, Google knocked $100bn off its market value after a promotional video showed its bot, Bard, sharing inaccurate information about discoveries from the James Webb Space Telescope.

While Google’s gaff is an extreme example, all businesses – no matter their size – must ensure full oversight of, and a structure of accountability for, AI in their organisations. As we discussed in a white paper published last year, there are ways companies can effectively govern use of AI, such as making it a standing item on the risk committee, and ensuring that the team responsible for building or implementing AI tools is separate from the team ensuring its ethical use.

“The simple fact is that this capability is now available for everyone, which makes it a virtually valueless commodity, as it adds no differentiation.” – Alexandre Pintos, SVP Data Science at Signal AI

Will ChatGPT wave in a new era for industry? While leaders, investors and consumers are certainly excited about its potential (there have been more than one hundred FT articles with ‘ChatGPT’ in the title since the beginning of this year), the technology has its limits. “The simple fact is that this capability is now available for everyone,” reflected Alexandre Pintos, “which makes it a virtually valueless commodity, as it adds no differentiation.”

The true impact of conversational AI tools in our world remains to be seen. But the rapid emergence and wide excitement around ChatGPT proves that no tech firm can afford to sit back: even Google – the ultimate disrupter – has been seriously disrupted.

Are you planning on using ChatGPT in your business? Please get in touch – I’d love to hear from you.

liana.osborne@thembsgroup.co.uk | @TheMBSGroup