Live data from Hacker News

A man spent a year in jail on murder charge that hinged on disputed AI evidence

theregister.com

111–120 of 211 posts

Re: A man spent a year in jail on murder charge that hinged on disputed AI evidence

#111

Earlier quoted context omitted.

A year? Curtis Flowers spent over 20 years on Death Row without being convicted of anything. A couple years ago he was set free! Why? Because of a true crime podcast called In The Dark. Seriously. Listen to season 2 and your mind will be blown. The “justice” system in this country is horrific. https://en.wikipedia.org/wiki/Curtis_Flowers?wprov=sfti1

I agree with your distress at the state of the justice system, but it doesn’t make sense to say that someone was on death row without having been convicted. The article you link to explains that his convictions were overturned. Although I would never rely on information in a Wikipedia article, you did link to it, so presumably you find it credible.

Every time he was “convicted”, the prosecution was found to be in blatant violation of laws on appeal. They intimidated witnesses, kept black people off the jury, and more. So no, he was never rightfully convicted! The prosecution broke so many laws, way more than Curtis Flowers ever did! And they just did it over and over again for 20 years. None of them have faced any consequences. They effectively imprisoned an innocent man for 20 years because they didn’t like him. And big surprise, Curtis is black.

Re: A man spent a year in jail on murder charge that hinged on disputed AI evidence

#112
I get that their system may sometimes need manual classification of signals but how is it possible to change the triangulated location? That is the core part of the technology and doesn't require ML to implement. Seems like it's all just Theranos style vapor.

Re: A man spent a year in jail on murder charge that hinged on disputed AI evidence

#113

The article says that employees of the AI company (ShotSpotter) manually reviewed and classified the sounds as gunshots: > records showed that ShotSpotter actually initially picked up what sounded like a firework a mile away, and this was later reclassified by ShotSpotter staff to be a gunshot at the intersection where and when Williams was seen on camera. So the AI didn’t even make the call. The staff did, manually.…

It's scary that prosecutors thought that the ShotSpotter evidence is suitable for a court. It's amazing technology that can help police get to a crime scene quickly but is far too easy to spoof or get the wrong answer due to reflections and other sounds etc. Hopefully cases like this will eventually stop the system being used in court.

My personal theory has long been that "technology" is the word people use for things that don't reliably work. It's never clean drinking water from the tap or the seed drill, it's stuff like this or clunky meeting software.

Re: A man spent a year in jail on murder charge that hinged on disputed AI evidence

#114
post #87

Earlier quoted context omitted.

https://theintercept.com/2019/05/05/forensic-evidence-aafs-j... https://www.rollingstone.com/politics/politics-features/jeff...

That doesn't exactly support your claims for most of forensics. If you read/skim the NAS report these articles are based on, you will see that much of the issue is not with the underlying science, but rather in the "interpretation" part. Yes, the odontology seems to be unsupported. But even the fire analysis that you list as an example is said to have a scientific basis like chemistry, but that the interpretation par…

While my stance on forensics is not as negative as the parent comment, I don't see how you are drawing such a sharp distinction between the "science" and the "interpretation". They are not separate things.

When a scientist proposes a new test, they include measurement techniques as well as a method of interpretation. If a medical test is used in a way that is frequently prone to misinterpretation, it would be fair to call that entire test useless.

Re: A man spent a year in jail on murder charge that hinged on disputed AI evidence

#115
post #9

Coming next year: "A man spent a year in jail on child pornography charge that hinged on disputed Apple CSAM evidence".

For that to happen: they'd have to be matched 30 times, the Apple employees would have to mistakenly manually confirm the match 30 times, and the defence would fail to request to see the underlying images that matched the hashes during discovery. It seems... far fetched.

You don't know it's 30 and you don't know it's going to stay so even if it is.

Re: A man spent a year in jail on murder charge that hinged on disputed AI evidence

#116
post #87

Earlier quoted context omitted.

https://theintercept.com/2019/05/05/forensic-evidence-aafs-j... https://www.rollingstone.com/politics/politics-features/jeff...

That doesn't exactly support your claims for most of forensics. If you read/skim the NAS report these articles are based on, you will see that much of the issue is not with the underlying science, but rather in the "interpretation" part. Yes, the odontology seems to be unsupported. But even the fire analysis that you list as an example is said to have a scientific basis like chemistry, but that the interpretation par…

It may very well be the case that a version of fire analysis that is scientifically rigorous can be done, but this would have nothing to do with how it is typically practiced.

Re: A man spent a year in jail on murder charge that hinged on disputed AI evidence

#117

The article says that employees of the AI company (ShotSpotter) manually reviewed and classified the sounds as gunshots: > records showed that ShotSpotter actually initially picked up what sounded like a firework a mile away, and this was later reclassified by ShotSpotter staff to be a gunshot at the intersection where and when Williams was seen on camera. So the AI didn’t even make the call. The staff did, manually.…

There was a submission here a few weeks back about ShitSpotter and how they seem willing to put in overrides at request of the police without necessarily having a basis for it.

If that's what happened in this circumstances it would be, police check ShoySpotter for more evidence, ShotSpotter hasn't automatically marked anything, police request manual review for gunshot they are sure must be there, ShotSpotter manual review process comes up with something and marks it as a shot with a human override.

There's a lot of places bias, or outright corruption, could cause problems in that process, if it's what happened here.

Re: A man spent a year in jail on murder charge that hinged on disputed AI evidence

#118
post #60

It seems to me jail time without verdict is the real problem in this scenario — it’s not clear to me how that’s not “guilty until proven innocent”

In Germany and Austria you get compensated for your time incarcerated if found innocent. So as to basically "fine" the government for not having speedy trials.

Re: A man spent a year in jail on murder charge that hinged on disputed AI evidence

#119

Not really relevant to this article, but relevant to the title: if AI keeps banning accounts for reasons that seems nonsense to humans, how will we protect people from getting flagged and locked up 'just in case' by the machine? AI is just not advanced enough to jail people, by a very large margin. And yet we will see it happen.

The scariest part is how quickly policing agencies are to embrace this tech as well but the government as a whole is slow to change anything else. Why is it that the state can be proactive about jailing me but not as much when it comes to quickly resolving cases and protecting my rights?

Re: A man spent a year in jail on murder charge that hinged on disputed AI evidence

#120

The article says that employees of the AI company (ShotSpotter) manually reviewed and classified the sounds as gunshots: > records showed that ShotSpotter actually initially picked up what sounded like a firework a mile away, and this was later reclassified by ShotSpotter staff to be a gunshot at the intersection where and when Williams was seen on camera. So the AI didn’t even make the call. The staff did, manually.…

It's the AI companies to blame, always. Obviously the algorithm is without blame, it's not a blamelable entity.
Post reply on HN