
22:18
Hi everyone!

22:24
Hi Camille! :)

22:24
Hello! Lovely to see you all!

22:24
hello from the university of Greenwich

22:25
I am Paul Mersh - PhD student at Greenwich

22:27
Hello!!

22:38
Hi!

22:45
Hey! UoG undergraduate

22:51
Hi everyone - I am from Manchester IP but I live in Sheffield -so hi from Sheffield

22:55
Hello from the US!

23:04
Kerstan-Student from US (UNC)

23:06
Hey!

23:14
OU law student. UK. Hi all

23:30
Hello everyone!! I’m Amalie from Norway and I study at the university of greenwich

23:38
Hi everyone, I’m a master’s student from the University of Greenwich

29:02
Hi all! ask any of your questions in the Q and A box using the button on the task bar at the bottom of the screen.

29:25
If any questions get lost in here, I will put them there... enjoy!

31:39
oops Forensic student here 🤭

44:25
yes they do and Mike O brien challenged it and the court said it was OK.

44:34
Remember to ask any questions for Brandon, Dean, and John in the Q and A

44:39
Great book and in our library at Manchester

45:56
what was the name of the book?

47:29
Unconscious bias perhaps?

48:07
Not so UN-conscious?

49:03
Autopsy of a Crime Lab Brandon Garrett’s book

49:58
This book explains the problem of expert testimony https://www.goodreads.com/book/show/114943.Whores_of_the_Court

50:23
thank you all ✨

54:50
Indeed -that was my point -NOT hearing it

56:43
So many reasons to bring about the end of the jury system.

57:20
it’s really interesting to hear how things were done with documentation and continuity in the past in comparison to what it is now.

01:00:05
Lee Morgan - see our work on https://www.justicebrd.com showing how machine learning may have a role here in the future.

01:02:11
...monster algorithms?

01:05:03
Re machine algorithms. Have read Abstract and looks good. Reminded of statistical vs clinical prediction (Paul Meehl).

01:08:25
https://www.ukessays.com/essays/psychology/clinical-judgement-vs-statistical-predictions-4887.php

01:08:35
Also, even though ML can indeed have a role, to what extent are we sure that it eliminates bias?

01:09:13
Can one ever eliminate bias? I think the best one can hope for is to limit its influence.

01:09:38
Bias in machine learning is exremely important and is covered in detail in our paper and sensitivity analysis.

01:10:26
https://academic.oup.com/lpr/advance-article-abstract/doi/10.1093/lpr/mgaa003/5807887

01:11:14
Thank you Cliff Mitchell.

01:11:21
Full article here: https://academic.oup.com/lpr/article/19/1/43/5807887?

01:12:41
'This model has been 'trained' on a number of murder cases where the true outcome is already know and then tested on many other known cases and shown to produce the correct verdict with 100% accuracy.' doesn't this assume the verdicts are correct in the first place? is this just repeating identified patterns?

01:15:01
The training cases are based on exonerations and 'true guilt'

01:15:28
Yes Alex, very well highlighted I think

01:15:55
Cliff Mitchell : do you have anything further on The naïve Bayes classifier?

01:16:24
Giah, exactly! Good point!

01:19:13
what is 'true guilt'?

01:19:33
Naive Bayes was chosen for its simplicity and proven to work with small data volumes. The paper has references to appropriate sources.

01:20:20
@Giah I don't know how to insert a heart in the zoom chat but <3

01:23:44
Thanks Cliff, am unable to follow links for some reason.

01:25:55
We use a number of criteria to define 'true guilt' or sound convictions including "...the defendant was found guilty by a jury and later the conviction was supported by, say, reliable confessions of guilt or guilty pleas at a re-trial"

01:28:44
Please just cut and paste this: https://academic.oup.com/lpr/article/19/1/43/5807887

01:33:39
YES- see the new report in the UK - BUT alas until it happens to you most of the public are not interested ...

01:33:45
@Claire I strongly agree with that

01:34:21
let us in the IPL know for sure :)

01:34:50
Thank you all

01:35:42
Thanks very much to all!

01:39:50
Please do ask questions of these wonderful students in the Q and A

01:42:47
You can check out the cases for the IPL at www.IPLondon.org

01:47:06
PechaKucha is a storytelling format where a presenter shows 20 slides for 20 seconds of commentary each. These were great for the IP this year

01:49:36
Amelia - 100%! So many opportunities and quite a lot of travelling too to symposiums which is so exciting

01:49:38
Amelia, lots of these students are undergraduates, I will ask your question in the 'questions'

01:50:35
A surgery?

02:11:24
We train students and support them -they also learn as they go.

02:13:24
So good to hear from you all and thanks for coming - WE CAN ALL PUSH ROCKS - even if it is not that far...

02:25:05
Thanks to all of you for your great questions and comments.

02:25:32
Thank you everyone. That was amazing! And you’re doing amazing work!

02:25:33
thanks to the speakers too, for sharing your experiences.

02:25:44
well done Megan for staying up all night !!

02:26:42
Thank you!

02:26:43
Thank you to the speakers, it’s been an incredibly inspiring evening!

02:26:43
Thank you everyone!