Cold Case Review of Sophie Tuscan du Plantier murder to proceed

he doesn't seem to have had any criminal/violent/recidivist history. or known substance issues.

not that this concludes anything. but it would just put him outside the typical pattern for this crime according to ai.

is that interesting? up to you.

Statistics apply to a population, not individual instances.

Back to talking about the gate so.

Lets recycle page four hundred. Again.

good point. its probably coincidence. profile shmofile.

now back to going round in circles.

you mean scutter in, scutter out as they say down in west cork

But AI is not using, "overarching established facts of the case". AI isn't magic, and it is isn't intelligent, it can only parse whatever data is scooped up online. It is not a human, it can not think, "hang on, this data doesn't pass the whiff test".

AI on this case would give a vastly weighted response in favour of Bailey being guilty exactly because of the amount of badly researched copy and paste articles there are online that it would use and give such a weighted average to its result.

From your comments, I don't believe that you know how AI works and why there are huge concerns about it from those that do.

If AI was to parse only this forum thread then it would likely conclude Bailey is not guilty, would you agree with that AI result?

I may have misunderstood, but I thought the original proposal was asking an AI source to create a "profile" of the likely criminal. Like "Profilers" do.

They don't generate a name, but rather a sort of sketch of the kind of person the criminal is likely to be; local or foreign, male or female, tall or short, sociable or reclusive, impulsive or very rational. Etc etc.

If, in this case, the AI bot profiled someone who is male, impulsive and has a history of violence - well, you might think that matches Ian Bailey.

It does, but it also matches at least half a dozen other men in the vicinity. Probably more. So it is, as Card pointed out, back to square one, with no £200.

The input isn't just the question / query it is also (and possibly more importantly) the dataset it has been trained on.

You're very mistaken and arguing against assumptions you have made yourself.

Depending on the prompt yes the ai will use the overarching established facts of the case to deliver an offender profile.

The task for the ai was to categorize the crime, and conclude a profile of the probable offender, not to name a name or to investigate clues within the case which indicate any particular person. (so your question about ai using just this thread to name a name is flawed as a comparison)

You will have done something similar yourself automatically, everyone naturally dismisses old ladies and toddlers from this crime. We may also conclude it probably wasn't a race crime, or a political crime. Logic which doesn't need analysis of alibis or witnesses, more macro than micro.

You can ask ai to take that same basic profiling which we humans do to the extreme with its massive artificial processing power, and you can ask it to not just regurgitate prior human conclusions, but rather to use its vast potential and ability to process information from many sources to correctly place this crime with others like it, and use similarities to point out the most likely characteristics of the offender.

In this case it dug into professional literature which I would never have thought to access, and explained correlations and why it was concluding its final delivered verdict.

Which by chance correlated closely with findings of 2 teams of professionals in various psychological disciplines.

Why you react badly to not only the human conclusions and the ai conclusions (two separate ais same conclusion btw) is the question you need to ask yourself.

None of them actually accuse anyone, no particular name is given. Nobody is put in the dock.

Just a profile of a personality type likely to have committed this crime is described.

Objectively when 2 ai's and 2 teams of professionals come to a similar conclusion on the profile there are good odds that it means something.

And if Im flat out of options, going round and round in circles on page 450, playing the same old record, like some in here, well I'd take it over nothing. Which is where you are. After many years.

AI isn't going to solve this crime.

This thread is a great example of stupid **** people do with AI while failing to understand it.

Welcome back. I don't recall a Will Thomas being mentioned before?