The frontend in AI is like tap for consuming water. As long as there’s a tap, people don’t generally think if the source of water is still clean. As long as they don’t have to fetch from the river, it all feels acceptable. AI has been powering products for decades, in the backend, but the moment it became visible is when the scepticism sparked higher.

This is one of the strongest challenges for AI product creators at the moment: keeping the users comfortable with the whole “black box” concept. Even though the water purification process in an area is black box too, the comfort level of people with it is high. How are you sure that the employees of the water management body are doing their job properly? Just because a pandemic isn’t there yet?

That’s where we are heading in AI. The comfort level. That’s all there is to it. Does that mean 5-10 years from now everyone would have accepted it? Of course not. Tap water isn’t either. There’ll be more “form factors” to claim how their AI is more pure and mineral rich. This is just the beginning. On a similar note, MIDI packs claiming to have artist-specific format of chords and patterns are widely and openly purchased, advertised and used. Not many people find that objectionable, biggest artists included. Who is to say those weren’t AI generated and sold in the past year? This is just like packaged-bottle, downloaded and served from an actual source that the user never gets to see. Law binds things together. But all it does in essence is provide comfort.

On a similar note, ADCx India has talks around audio analysis with ML and generative AI in music. Attend here: https://audio.dev/adcx-india-24

PS: 20 years ago many people - particularly government employees who usually don’t put effort learning anything new - were sceptical of using computers and would prefer to make official data entries on paper. Many of those records are unavailable now because of them. Just saying.