Appreciating Human Flaws In Reaction To AI

The year is 1493, and Christopher Columbus returned to Spain after his first expedition. He returns with many goods, but mainly a little gold. This gold started numerous Conquistadores' attacks on the Native Americans. Still, the gold they brought back to Spain profoundly impacted their economy, as if an Inca curse befell the country. The main effect was inflation because gold lost purchasing power as it became more plentiful. The law of offer and demand applies to many fields, and a few weeks ago, I explored in another article how conciseness may lose value as AI can quickly summarise content. This week, I noticed how Microsoft Outlook started highlighting unnecessary words in the emails I was typing. It did not take long for my prediction to become true.

You will find many authors who have analysed the traits and habits of the most successful people. AI experts seeking the best performance would apply as much as possible to their various models mimicking human behaviour. It could be this brunette with long hair in a red dress playing the piano, which you might have seen in a video on LinkedIn recently. As AI simulations of human behaviour seek to be as perfect as possible, this perfection makes me spot them as AI content immediately. However impressive in realism, we might start valuing slight imperfections and quirkiness as signs of genuine human behaviour.

The appreciation of imperfections and even seeing beauty in them is a concept that the Japanese know well, called?wabi-sabi?and a closely linked tradition called?kintsugi. Kintsugi is repairing broken pottery pieces, such as bowls and tea cups, instead of throwing them away. The word means golden joinery, which uses lacquer with either silver, platinum or even gold. The result is so beautiful that it makes you want to smash your bowls as soon as you return from the shops, just so that you can start the repair process. I would not be so surprised to see this appreciation and people finding beauty in our behaviour's imperfections in response to AI's rise.

I wonder how AI creators would respond to a rise in AI-driven wabi-sabi. Adding an element of randomness to AI's errors is certainly possible, but the result risks missing the mark and even induces uncertainty bias in the audience. People would dismiss and fail to engage in the experience fully, only to fall back on the older but time-tested alternatives. It is probably a safe bet to assume that AI visual experiences will continue erring on the side of looking a little too perfect at the expense of realism. However, there is the uncanny demo unveiled at the Google I/O Conference in 2018 for purely audio AI content. Google Duplex made a hair salon booking appointment full of natural-sounding pauses, "um"s and "mhh-mmm"s.

When AI content is almost indistinguishable from genuine content, some people report feeling uneasy. Japanese roboticist Masahiro Mori coined the term uncanny valley as early as 1970. The early robots were not human-looking, but the ambition to build robots that looked as human-like as possible became more apparent over time. There is a moment when the robots look human but are still imperfect. Up until that point, the robots were a little more than showing the art of the possible. But with increased realism, people will start wondering whether they could or should depend on them. Reaching the other side of Uncanny Valley will take tremendous effort, and we might see the manifestation of the 80%-20% rule, or Pareto Principle, where the last 20% of effort takes up most of the time and resources.

For more articles like these, follow me on LinkedIn or X: @albangerome

#MeasureCamp #AI #WabiSabi #Kintusgi #UncannyValley #WAWCPH #CBUSWAW