Does it matter if ChatGPT produces Bullshit?

Many users report that for many questions, ChatCPT produces plausible bullshit — sounds right but wrong — just like the average BS artist you walk away from.

Professor Harry Frankfurt wrote a paper on the subject in 2005 which Professor Gary Marcus refers to in a recent Ezra Klein show, “The essence of bullshit is not that it is false but that it is phony. In order to appreciate this distinction, one must recognize that a fake or a phony need not be in any respect, apart from authenticity itself, inferior to the real thing. What is not genuine may not also be defective in some other way. It may be, after all, an exact copy. What is wrong with a counterfeit is not what it is like, but how it was made.”

To this end shortly after its release, Stackoverflow banned the use of ChatGPT-generated code as it had so many errors.

The real problem is that the cost of generating misinformation decreases to zero while the cost of valid information remains the same. The audience for catastrophes is catered by a media that seeks to maximise profits. The media will be increasingly challenged to publish validated stories in a sea of misinformation that sounds plausible but should you take the time to review all references simply carefully constructed bullshit.

So we know we can be fooled by a plausible pitch and the rush to Artificial General Intelligence (AGI) which in 2015 was detailed as having a 10% chance of being realised by 2022 and a median expectation of achieving AGI by 2040 — this is now closer than anticipated only seven years ago.

Artificial Super Intelligence (ASI) follows 20 years later when computers can achieve a performance of many orders of magnitude that of humans — the article assesses this to have a 50:50 chance of being beneficial to humans — not being beneficial is an extinction event as per all the dystopian movies we love and after which this blog is named.

The good news is the “shiny new thing” that is ChatGPT is brilliant but imperfect — like most new things. ChatGPT may produce plausible bullshit today this is simply an opportunity for improvement.?

Microsoft who has invested over $1b already are reportedly going to integrate ChatGPT into their Bing search engine in the next few months. The response from Google was to declare a “Code Red”, which is evidence that we are at the beginning of a new industrial revolution potentially more impactful than the Internet thirty years ago.

Realists like Gary Marcus, who promote a pragmatic approach to AI adoption help us achieve balance. I am currently reading his book In the 2019 book “Rebooting A.I.: Building Artificial Intelligence We Can Trust” Marcus and his co-author Ernest Davis outline a path to A.I. development built on a very different understanding of what intelligence is and the kinds of systems required to develop that intelligence.?

So think about what is necessary rather than nice to have and consider the evolution of products over the last 10 years — Skype, Viber, WhatsApp, Signal and consider your adoption — you used what is of most value in terms of service and usability. I always like to think of this as heuristic adoption — my innate laziness drives my preference.

In my last article, I detailed the risks of becoming an expert dodo — this is an easily avoided extinction opportunity.

I am open to change and in my consulting business help others see perceived threats as opportunities — resistance is futile is an otherwise reality.

Don't excuse your resistance by blaming the current imperfect nature of the ChatGDP tool — if, as the joke goes, bullshit will get you to the top but won’t let you stay there, don't use it as an excuse to miss the party.