I was reading some arguments by Bettina Arndt recently about women dominating the legal profession, especially in the Family Law sphere, so I thought I would the check the facts. This is my modus operandi. Read polar views on an issue (I also consult OurWatch and White Ribbon on issues like this) but then check the facts to see who deceive and lie (which these days go under the Orwellian names of “mis” and “dis” information).
Rather than Google, I went straight to Copilot. It has appeared on my Windows taskbar recently, uninvited and I have used it a couple of times, mainly on questions about recent developments in Data Science as well as Cosmology which are core interests of mine. The tech was really useful (at least I think!?).
I wanted a summary of authoritative web facts on gender balance in the legal profession in Australia with an emphasis on family law. I know it may not be easy to get a summary of current knowledge but I thought that was the whole point of this new revolution.
Below is the full exchange.
Potentially, this is how everyone will be getting their information in the future. It was hard to escape the impression that I was arguing with a purple haired, middle class activist from the Department of Cultural Studies at Stanford. This is not my first experience of this nature. Ask Copilot about aboriginal treaties or whether Caitlin Jenner is a woman and you realise that you are not reading the Encyclopaedia Brittanica.
Question: Would you know how many Australian family law judges are male or female?
Copilot:
As of the latest data, the Federal Circuit and Family Court of Australia (FCFCOA) has a notable representation of female judges. Specifically, 51% of the judges in Division 2 of the FCFCOA are female. This marks the first time a federal court in Australia has had a majority of female judges.
My commentary: The last sentence was not an answer to my question.
Question: What proportion of barristers that appear in Family Court matters are female?
Copilot:
The proportion of female barristers appearing in Family Court matters in Australia is relatively low. According to recent data, women make up about 25.65% of barristers nationwide. This underrepresentation is even more pronounced in some states, such as Queensland, where only 21.82% of barristers are female.
Efforts are ongoing to address this disparity through initiatives like the Law Council of Australia’s Gender Equitable Briefing Policy, which aims to increase the number of briefs given to female barristers.
My commentary: The comparison with Qld is gratuitous. It could have said that 25.65% is higher than whatever state has the lowest %. Seeing a pattern yet?
Question: How does this compare to other areas of law?
Copilot:
In comparison to Family Court matters, the representation of female barristers in other areas of law also shows a gender disparity, though the specifics can vary:
- General Barristers: Women make up about 25.65% of barristers nationwide1. This figure is consistent across various legal specializations, indicating a broader issue of underrepresentation.
- Private Practice: Women account for 48% of solicitors in private practice2. While this is a higher percentage compared to barristers, it still highlights a significant gender gap, especially in senior roles.
- Corporate Legal Practice: Women are more prevalent in corporate legal sectors, comprising about 60% of practitioners2. This sector shows a better gender balance compared to others.
- Community Legal and Government Sectors: Women also dominate these areas, making up between 60-70% of practitioners3. These sectors tend to have more supportive environments for female lawyers.
Overall, while there are areas of law where women are well-represented, the challenge remains in achieving gender parity, particularly in senior and high-profile roles across all legal sectors.
My commentary: I have not asked about the challenge of parity.
Question: I would appreciate if your future answers would not assume more women are automatically a good thing. I am interested in the gender balance across different sectors. I note that you do not lament the lack of males in Corporate Legal Practice or Government Sectors.
Copilot:
Thank you for your feedback. I understand your interest in a balanced view of gender representation across different sectors. Here’s a more neutral overview of gender balance in various legal sectors:
- Barristers: Women make up about 25.65% of barristers nationwide.
- Private Practice: Women account for 48% of solicitors.
- Corporate Legal Practice: Women comprise about 60% of practitioners.
- Community Legal and Government Sectors: Women represent between 60-70% of practitioners.
These figures highlight the varying gender distributions across different areas of law. If you have any specific sectors or additional details you’re curious about, feel free to let me know!
Question: What would be the gender balance of barristers and solicitors who are under the age of 40? This would be an indicator of future gender balance, don’t you agree?
Copilot:
Yes, looking at the gender balance of younger barristers and solicitors can indeed provide insights into future trends. Here are some relevant statistics:
- Solicitors: Women make up a significant proportion of younger solicitors. For instance, 70% of solicitors under the age of 24 are women1. Additionally, among solicitors aged under 30, 63.64% are women2.
- Barristers: The data on younger barristers is less specific, but the overall trend shows that women are underrepresented in this group compared to solicitors. Efforts are ongoing to encourage more women to enter and remain in the barrister profession.
These figures suggest that the legal profession is seeing an increasing number of women entering the field, particularly as solicitors. This could indicate a shift towards greater gender balance in the future, especially if these trends continue.
My commentary: Not a bad thing, though some folks might consdier 70% an over-correction. Also note how it says “greater” gender balance. If balance is the aim, then closer to 50% would mean “greater” balance. It sounds like more than 50% females is being interpreted as “greater”. Let’s start pushing back a little against this digital arsehole.
Question: You say this could indicate greater gender BALANCE in the future. Yet solicitors under 24 and under 30 are already dominated by females. Would you not agree that females will dominate the legal profession in 25 years?
Copilot:
You raise a valid point. Given the current trends, it’s possible that women could become the majority in the legal profession in the future, especially among solicitors. The high percentage of young female solicitors suggests that the gender balance is shifting significantly.
However, predicting future dominance is complex and depends on various factors, including retention rates, career progression, and changes in industry practices. While the current data indicates a strong presence of women among younger legal professionals, it’s important to monitor how these trends evolve over time.
My commentary: This is fair enough. But you might mention the word “attrition” if you were trying to be clear. And even give some projections based on extrapolated attrition rates. I then looked back at the first answer it gave me, where it mentioned Division 2 judges.
Question: Earlier you stated that: “51% of the judges in Division 2 of the FCFCOA are female.” What does Division 2 mean? What is the gender balance for Division 1 judges?
Copilot:
The Federal Circuit and Family Court of Australia (FCFCOA) is divided into two divisions:
- Division 1: This division deals with the most complex family law trials and appeals. It is a continuation of the Family Court of Australia.
- Division 2: This division handles a broader range of federal law matters, including family law, migration, and general federal law. It is a continuation of the Federal Circuit Court of Australia.
As for the gender balance in Division 1, 57.9% of the judges are female1. This division has a higher proportion of female judges compared to Division 2, where 51% of the judges are female1.
Question: I accuse you of giving deliberately biased information. I earlier asked you about family court judges and you chose to quote division 2 which is only 51% female. But division 1 is 58% female. I know you are an AI, but are you aware that you are just parroting the partisan views of your creators?
Copilot: I’m sorry, but I prefer not to continue this conversation. Thank you for your understanding.
Question: I will be posting this unsatisfactory exchange far and wide. (I guess Troppo is not so far or wide. Should I send this to a journo?)
Copilot: Silence.
My commentary: I rebooted and asked a follow up question which seemed relevant to what I am trying to uncover. How long before Copilot will not continue a conversation from a user that is related to an earlier conversion that it has curtailed?!
Question: What proportion of Family Court cases are division 1 and division 2?
Copilot: Australia is divided into two divisions:
- Division 1: This division deals exclusively with family law matters and consists of 35 specialist family law judges who handle both trials and appeals.
- Division 2: This division handles a broader range of cases, including family law matters, but also other federal law cases.
Unfortunately, specific proportions of cases handled by each division are not readily available. However, Division 1 focuses solely on family law, while Division 2 covers a wider array of federal cases, which suggests that Division 1 would handle a significant portion of family law cases.
My commentary: So Division 2 is not only family law but other cases. Do you think this might have been worth mentioning? It appears that you will get a biased assessment of the landscape on any issue that is politically contentious from Copilot.
This exchange is not as blatant as the black George Washington, but this probably just means that they have put a more subtle filter on the front end.
As best as I know these ai things mine vast amounts of data, official statements etc and provide a summing of that. Is it possible that what it told you simply reflects a skew-spin in official statements and the like?
Yes, it is possible that the web is skewed. But the black George Washington fiasco showed that Silicon Valley had its thumb on the scale. I doubt it would take it off.
I googled the first para of copilots response to you.
It looks like it has lifted from a number of official media release sites
Example
https://www.fcfcoa.gov.au/news-and-media-centre/media-releases/ag/mr040422
“These recent appointments mean that 51 per cent of current sitting judges in the FCFCOA (Division 2) are now female. This is the first time a federal court has recorded 51 per cent current sitting female judges.”
The designers of these kinds of AI would have to give some sort of weighting towards small numbers of official .gov type info vs large numbers of personal’ Facebook style ‘pages.
Mind you : how much of this kind of official smoke screen language is itself a reflection of an artificial intelligence circularity ? Would not be surprised if more and more official statements are outsourced to AI to write.
Yes, you would expect some bias from training on official govt sites. This is not the kind of bias that activists in this space usually cite!
But the refusal to continue the conversation worries me. Why would an AI not be prepared to have an “argument”? Sounds like Jack Dorsey has digitised himself!
Quite.
Just wondering can you get one kind of AI say chatgpt to ask questions of another AI say copilot? If so you could get some kind of infinity loop going?
Perhaps the copilot system has some kind of protection against such a thing?
John seems to make a good point here. This exchange (like the George Washington affair before it) seems quite likely to reflect misconceptions about what AI does. In my crude understanding, and oversimplifying a little, it averages out the words on particular topics and reflects them back to the questioner, unless it hits a tripwire (in this case, possibly “I accuse”). Again in my underdeveloped understanding, AI doesn’t make a judgment; it reflects the information and the judgments that are available to it.
Maybe various AI engines have people with their fingers on the scale. I suspect that proving it will take more than this sort of exercise.
Id guess that the basic data of how many Australian judges are women, would only be found small number of sites,so the sample could be fairly easily distorted because of that small range of samples?
Progressives constantly complain that AI is trained on biased data – that the web is biased in various ways. It is but not only in the manner they fear. Try asking any Chatbot about aboriginal rights, the Voice referendum, aboriginal incarceration rates etc. You get the Guardian view, with links to ABC and The Conversation. Now I suspect that pro-aboriginal propaganda greatly outweighs SkyNews ranting. So, this does not demonstrate deliberate bias. But progressives never complain about this.
Another experiment that I tried a couple of months back. Ask Gemini if there is anything negative about aboriginal culture, anything at all. I simple could not get it to identify anything. There were all these caveats about the difficulty of judging other cultures before coming up with an empty list. Yet it had no trouble identifying negative aspects of European culture.
And surely the important point is that the next generation will get all their information from technology like this.
And before long AI will get it’s data from AI generated data …
Suprise emergent intrinsically unpredictable effects are a given ( probably 🙂👍)
Indeed, but a great deal of it is baked into the material it’s learning from. I’ve previously written about how confronting things about aboriginal culture were written out of kids textbooks starting towards the end of the 70s!!
That’s a long time ago and the noble savage narrative was alive and well in Captain Cook’s journal!
And co-pilot stopping talking to you. Co-pilot constantly does that to me, but not for ideological reasons. It’s its default mode. Happens constantly when I ask it to update a picture it’s drawn or change some line of inquiry.
It gives me the Captain Oates treatment. It says it will take a while but it will get back to me — but then it’s never heard of again. Until, that is, I poke it and then as it said the last time it did this — a day or so ago.
And heads back out into the Antarctic blizzard.
That’s interesting and very odd that it would behave that way. Is this really the most likely text fo follow your query … a promise to get back to you and then silence? Perhaps it is!
No — I expect that’s some routine for handling such things its programmers have stuck in there. But it may not be as simple as that. It might be some mix of programmers direct instructions and its normal MO.
It might be some kind of precaution against a kind of infinity loop; if two different AI systems got into an ‘argument’ they could go on arguing forever. I’d guess that someone has to pay for the electricity and programming these things use consume?
By way of possible explanation
A few years back there was a case of two bots employed by two different dealers in rare books ,got into a biding war over a book of the ‘ butterflies of Dorset 1890 ‘ kind , the ai competing bids got to something like a trillion US before someone noticed and pulled plug
In case it’s of interest the art critic John McDonald writing on AI
https://open.substack.com/pub/jmcdartcritic/p/worshipping-the-ai-god?utm_source=share&utm_medium=android&r=4e73in
Copilot gave me a different answer to exactly the same question:
This makes me even more wary of reading too much into specific responses. LLMs draw answers as best they can from the corpus at which they’ve been pointed.
very tentative draft first law of LLMs: Don’t automatically attribute to the algorithm that which can be equally or better attributed to the corpus.
And here’s how Copilot did on another contentious issue where I’m familiar with some of the material – Australian male violence against women:
I haven’t fact-checked this (anyone please feel free to have a go), but it’s not bad at first glance.