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

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

theregister.com

Post reply on HN
Print view View original

181–190 of 211 posts

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

#181

Post by Spooky23 »

giantg2 wrote: post #136
Spooky23 wrote: post #106

Earlier quoted context omitted.

I don’t know about that. Judges have front row seats to the 10,000 ways that cops, prosecutors and defense attorneys screw up every day. If anything, I think the expense of defense and the poor quality of public defenders probably accounts for bias.

Magistrates aren't even required to pass the bar. I had one think that I was calling him prejudiced when asking to dismiss with prejudice (and many other examples with other magistrates, but this one is somewhat funny). I had a judge say that you can't use statements made at a prior trial to discredit the same witness at the current trial if their testimony conflicts. He also denied said that a trial de novo is a "co…

That’s bullshit. Lay magistrates deal with minor issues like traffic infractions, violations and minor misdemeanors.

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

#182

Post by giantg2 »

Spooky23 wrote: post #181
giantg2 wrote: post #136

Earlier quoted context omitted.

Magistrates aren't even required to pass the bar. I had one think that I was calling him prejudiced when asking to dismiss with prejudice (and many other examples with other magistrates, but this one is somewhat funny). I had a judge say that you can't use statements made at a prior trial to discredit the same witness at the current trial if their testimony conflicts. He also denied said that a trial de novo is a "co…

That’s bullshit. Lay magistrates deal with minor issues like traffic infractions, violations and minor misdemeanors.

Misdemeanors can keep people from getting a job. There can be pretrial restrictions, including custody. They also tend to be the ones having preliminary hearings and can issue warrants, even for felonies.

These are not just traffic violations and minor issues. And even if they were, that doesn't mean that people's rights stop mattering, nor is that a valid excuse for gross incompetence.

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

#183

Post by c_weighted »

kevin_thibedeau wrote: post #180
c_weighted wrote: post #169

Earlier quoted context omitted.

Locating on an echo at one more receivers rather than a perhaps quieter more direct path. Putting the wrong impulses together when computing a location. Computing the wrong start time for one or more impulses. If you want to read up on all the ways this kind of thing can go wrong, the technical term here is "source localization."

The military has no problem detecting gunfire direction and estimating range from a single microphone array on a mast. Shotspotter is a distributed network of multiple sensors across a wide area. If they can't make it work after 15+ years then they are hopeless.

Boomerang is colocated with the target. Multipath is less likely to be an issue because there is likely to be line of sight to the source. Selecting the right impulses is easier because the spacing of the microphones provides some tight timing constraints. You’re not likely to see inconsistencies in timing the impulses lead to location error because any effects of the environment are likely to be consistent across the pulses. Finally, the location accuracy requirements are far different. ShotSpotter needs to put a dot on a map accurate enough to find shell casings or a victim in a wide geographic area. Boomerang just needs to give a good bearing from the receivers to return fire. The range doesn’t have to be all that precise, and the mathematics suggests it isn’t.

These are two very different approaches to solve different problems.

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

#184

Post by giantg2 »

shkkmo wrote: post #152
giantg2 wrote: post #44

Earlier quoted context omitted.

I generally agree, but wouldn't necessarily say a lot of forensics is junk science. The science is generally sound (procedures and theory). Usually it's the prosecutors that present the forensics in a way that misrepresents what the results actually mean. Or lab techs make mistakes or perform misconduct (How to Fix a Drug Scandal on Netflix is interesting).

Not all forensics disciplins are created equal and some of them, such as hair forensics, are definitely junk science.

How is hair analysis junk science? Maybe you just mean that people misuse the facts to come to incorrect conclusions?

It's only supposed to be used to include or exclude someone, and based on the numbers/probabilities it's generally only useful for excluding someone. Of course that's not how most prosecutors use it...

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

#185

Post by TheOtherHobbes »

shkkmo wrote: post #155
giantg2 wrote: post #87

Earlier quoted context omitted.

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…

If you are using forensics in a trial, the validity of the "interpretation" part is key to it's use. The reliability of these interpretations has been historically hugely overstated. Given this history it is absolutely reasonable to reject most forensic evidence unless there actual good science supporting the reliability of expert interpretations.

The usual example is DNA evidence. The two most obvious fallacies are actually called The Prosecutor Fallacy and The Defendant Fallacy because it's so easy for either side to mis-sell what the evidence really means.

It's basically impossible for someone who isn't pretty handy with Bayes to work out the significance for themselves.

The average jury member has no chance.

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

#186

Post by giantg2 »

alisonkisk wrote: post #171
giantg2 wrote: post #44

Earlier quoted context omitted.

I generally agree, but wouldn't necessarily say a lot of forensics is junk science. The science is generally sound (procedures and theory). Usually it's the prosecutors that present the forensics in a way that misrepresents what the results actually mean. Or lab techs make mistakes or perform misconduct (How to Fix a Drug Scandal on Netflix is interesting).

Citation strongly needed on forensics bring generally sound.

There have been plenty of references in this post to the NAS study on the state of forensics.

Basically, the majority of forensics are based on sound theories and practices. The issues are when professionals and prosecutors try to "interpret" the facts.

I would be interested in any citations showing that forensics are generally unsound.

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

#187

Post by bagacrap »

MeinBlutIstBlau wrote: post #118
setr wrote: 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.

that sounds fair except not great that it's just penalizing the taxpayer, not anyone in government

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

#188

Post by DSMan195276 »

giantg2 wrote: post #93
BoorishBears wrote: post #89

Earlier quoted context omitted.

Because other surfaces are not a purpose built sensor designed to detect the features of a finger then repeatedly trained with a specific set of finger scans? Like are you joking? Saying TouchID validates fingerprinting is like saying MRI machines validate psychics.

Are you joking? It seems you two have a different idea of what is being discussed. The underlying fundamentals of finger prints are solid. The part that can make them ineffective is when they are using insufficiently tested tools or matching on too few points. And of course misrepresentating what a match actually means and how it pertains to the case.

> The underlying fundamentals of finger prints are solid.

What exactly are these underlying 'fundamentals'? Nobody even knows if everybody has unique fingerprints, it's just an assumption.

Beyond that, it's not all that relevant since as you mentioned how the matching is done is largely subjective, and it depends a lot on the number of "points" used and who does the matching, and that should really bring the accuracy even more into question. What exactly is the statistical probability of an incorrect match based on the number of points used? Good luck answering that question.

IMO the situation is made a lot worse by the fact that the public generally assumes fingerprinting to be extremely accurate (or 100% accurate).

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

#189

Post by smsm42 »

PragmaticPulp wrote: post #3

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.…

But this is so convenient! "Your honor, this is the AI algorithm that said it was a gunshot, AI can't lie and is based on pure Science!" Yeah, later it turns out people made that decision because police asked them to do it, but the defense may not know it.

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

#190

Post by MeinBlutIstBlau »

emodendroket wrote: post #168
MeinBlutIstBlau wrote: post #121

Earlier quoted context omitted.

The only time I hear about people who talk about AI are average joes, ignorant journalists, generic business analysts, and upper management that don't know anything about programming. I wish I could point to these people and say "look at video game AI, we can't even make those without it cheating or following a routine to simulate human playstyles. What makes you think we can do that in real life without doing the sa…

Video games are a bad example because it’s pretty feasible to make an AI that would win every time.

They are a perfect example because of that. They are in controlled environments. Life is the testing ground of everything. To think we can trust a computer to have reasoning skills in our environment ethically is absurd and a total affront to freedom.
Post reply on HN