[amluto]

I don’t think separating the disgusting mess in mixed recycling is an ML/robotics problem. Neither robots nor human hands can turn food-splattered paper into clean recyclable paper. Nor can they turn plastic/paper composite materials into anything else. Or dirty plastic bags or dirty polystyrene foam or (yuck!) fluorinated HDPE.

[cfr2023]

Fair observations but, doesn't this assume that disposal and containment are truly our only options now and forever?

Even so, you think there would be no worthy efficiency gains for pure sorting or materials mining in using machines that can identify known objects effectively, and hands that will go where humans will not?

[amluto]

There are solutions that don’t involve reaching into the mess and sorting it.

Some places incinerate their waste for energy. If done well, pollution (except CO2) is minimal, and the outputs are energy and ash. Perhaps some day the ash could be processed to extract useful minerals.

More generally, if you imagine waste to be a clumpy soup of organic goo, assorted interesting elements and minerals, and polymers (which are technically organic goo but are sort of worthy of a different category), perhaps it could be treated as such and processes could be developed to economically extract useful things from it, kind of like how geological processes turn dead cells into oil.

But I see no fundamental reason that, say, old polyester needs to turn into new polyester as opposed to anything else. And keep in mind that, even if burnt to CO2 and ash, there are processes that use energy that turn CO2 into valuable chemical feedstock.

[cfr2023]

Much to consider here, thanks for the thoughtful reply.