Smart, smarter, smartest

I pride myself on being clever and having a good memory. According to a test I took in college, my IQ is somewhere around 120. Not a genius, clearly, but certainly not bad. Recyclers tell me they are keen to make their operations even smarter with the help of artificial intelligence (AI).

A lot of things can be measured. Recycling rates, output purity, the speed of sorting systems, the percentage of accepted/rejected fractions, how many interruptions occur per machine – you name it. Having hard data at our fingertips, running a line of machines that ‘talk’ to each other and being able to troubleshoot from a distance boosts the confidence of the operators. It feels like there’s nothing they can’t control or fix, if they really need to.   

Besides the obvious drawback of major capital investment, and some hesitation as to which tech provider offers the best solution, the prospect of installing next generation sorting systems typically generates great enthusiasm. Recyclers want to show they mean business and they’re eager to be recognised as frontrunners, not followers.

I get it. It is a highly competitive market. Recovering 75, 85% or 95% of materials can make all the difference. And since AI systems are self-learning, their true potential evolves with every item picked and scanned.

According to UK’s Recycleye ceo Victor Dewulf, we’re already witnessing the ‘tipping point’ of AI-driven tech being deployed in the recycling sector. ‘It’s no longer a topic about the future. Change is here, now,’ he says.

This doesn’t mean that AI sorting is a perfect solution, an observation echoed by AMP Robotics’ founder Matanya Horowitz. ‘Even smart sorters have a learning curve. They can’t read your mind. They deliver reliable and fast results if they know what they’re looking at. Do you want to switch it up and treat multiple streams? Not a problem, it just takes some time to calibrate the system.’

Another area where AI is emerging as a resourceful tool is content creation. I have been playing around with Chat GPT and various audio-to-text programmes to see if they could enhance my work, even possibly yield some new ideas. In that respect, however, this AI is far from mature.

It was like talking to a drunk version of Wikipedia. An interesting experience but also rather frustrating due to the static and repetitious answers, limited data and plenty of mistakes.

Trying my luck converting an audio file into a useful transcript for interviews didn’t pan out either. Turns out, the human ear is much more skilled at adjusting to accents. With German, French or Indian, there is a certain flow to the way different languages are delivered. I know that as a journalist having covered international recycling themes for 11 years.

I laughed when reading from my transcription how Hensel Recycling was bothered by ‘bad teeth’ (thieves), how Tomra Sorting’s new test centre has ghosts (still not sure where that comes from) and how Professor Michael Braungart thinks banning straws is like rearranging ‘duck hairs’ (deck chairs) on the Titanic.

It shows that any new technology always comes with a margin of error. For me, it proves the power of our own creativity, critical thinking and intuition is invaluable.