Trust in AI in a post-truth world, is it possible?

Kim Nilsson
5 min readJan 17, 2023

My goodness has there been a lot of hype around ChatGPT in the last few weeks. And most of it is warranted — it is a truly remarkable piece of engineering. Yet, the tag line for the IG Nobel prizes comes to mind; it is something that first makes you laugh, and then makes you think?

Many of the first posts that came out from users of ChatGPT were hilarious. My favourite is still the description of how to remove a peanut butter sandwich from your VCR. It is easy to laugh at these earnest replies to what a human would consider a ridiculous question, and why not make us laugh? It brings joy to our lives.

Image: Pixabay

As I dug a bit deeper to understand potential use cases, I was at first ecstatic about the efficiencies a tool like this can offer. It is clear how it can compete with search engines and save a user valuable time. As a start-up founder, I often need to find market sizes for specific products of services. That would usually entail hours of googling, report reading, combining and synthesising. I asked ChatGPT for the market size for AI products and services globally, and within a few seconds it promptly told me it was $190 Bn in 2021. An exercise that would have taken me probably 4–5 hours was done in 5 seconds.

Other use cases further highlight the disruptive power ChatGPT can have on many white collar professions, especially within creative professions such as writing, journalism, and research. But what was initially a source of awe and enthusiasm for me, quickly turned to a source of concern and fear. The black box nature of these tools mean that it is impossible to lead any reply, any information, back to a specific source. Can we trust its outputs?

Some would say we are already in a “post-truth world”. With organisations and actors with a political agenda deliberately spreading false information, all the while accusing mainstream media and others of reporting “fake news”, and with troll farms in rogue states churning out misleading propaganda at a ruthless pace, judging what information online or in print is true is already made more difficult than it was only two decades ago.

Trust in mass media in the US has fallen from an all-time high of 72% in the late 1970’s to an all-time low of 34% in 2022 (Gallup), with a whopping 38% of

--

--

Kim Nilsson

Ex-astronomer turned serial entrepreneur. Founder, Mentor, Thought Leader in AI and start-ups. Writing about the the things I care about. Host of @FoundersYarn