Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

Walmart employees are out to show its anti-shoplifting AI doesn’t work

arstechnica.com

Post reply on HN
Print view View original

11–20 of 102 posts

Re: Walmart employees are out to show its anti-shoplifting AI doesn’t work

#11

Post by SkyBelow »

There is a lot of information in the article but little to draw a conclusion from. They bring up examples where the AI failed, but that alone does not indicate it doesn't work.

An AI that fails to recognize some thefts but does recognize others can still be considered to work as long as enough thefts are caught. Imperfect, but given the low cost of a failure relative to not using an AI, imperfection seems quite tolerable. Even the false positives seem to be of little concern as they involve a shopping associate coming over and having to assist an individual. As long as the experience for the shopper remains more pleasant than waiting in line for one of the cashier operated checkouts, it is still an overall win.

So yes, the AI is imperfect, but where is the information that justifies the conclusion it doesn't work and not the conclusion that it doesn't work perfectly?

Re: Walmart employees are out to show its anti-shoplifting AI doesn’t work

#12

Post by surething123 »

sudosysgen wrote: post #8
nojito wrote: post #5

False Positives don't matter. It's all about mitigating False negatives enough to lower shrinkage and justify the costs of the software.

False positives may cost more than the items being protected in diverse costs.

Yep. It may only take one or two false positives to lose a customer worth potentially orders of magnitude more than any particular item (depending on the price of the item, of course).

Re: Walmart employees are out to show its anti-shoplifting AI doesn’t work

#16

Post by code4tee »

This is a very hard problem and thinking a product could do this without extremely high error rates is naive.

Any implementation should use this as a supplement only, not replacement for humans. For example a human watching 10 checkouts and quickly manually reviewing scenes flagged as suspicious. Any implementation should also account for the fact that loss will still occur and account for that in the RoI.

The fact that the technology is very error prone is not surprising (to anyone that actually builds AI tech). The issue here sounds like poor implementation and unrealistic expectations from management on what AI truly can and can’t do.

Re: Walmart employees are out to show its anti-shoplifting AI doesn’t work

#17

Post by nojito »

sudosysgen wrote: post #8
nojito wrote: post #5

False Positives don't matter. It's all about mitigating False negatives enough to lower shrinkage and justify the costs of the software.

False positives may cost more than the items being protected in diverse costs.

Walmart's in-store revenue has been growing at a steady clip.

Re: Walmart employees are out to show its anti-shoplifting AI doesn’t work

#20

Post by microcolonel »

SkyBelow wrote: post #11

There is a lot of information in the article but little to draw a conclusion from. They bring up examples where the AI failed, but that alone does not indicate it doesn't work. An AI that fails to recognize some thefts but does recognize others can still be considered to work as long as enough thefts are caught. Imperfect, but given the low cost of a failure relative to not using an AI, imperfection seems quite toler…

Probably the reasons for attempting to undermine it are more political than practical or ethical.
Post reply on HN