Chat GPT and other AI bots

[Febbos]

"Summarize the content on NOT 606 transfer thread (link) and who thinks what in general"



Obvious that it took a liking to you Stockholm Tiger

You must log in or register to see images

You must log in or register to see images

 

[Plum]

I struggle to believe anyone's using it for legal letter or essays to any meaningful standard. Some of the responses it produces are just bollocks.

It's a step up from a search engine, if you know what you're looking for you can quickly separate the wheat from the chaff, but people using it blindly like that will soon get caught out.

My wife has to assess PhD theses, she says you can tell when stuff's come out of AI but it's getting harder and better from one term to the next.

 

[dennisboothstash]

Not personally, but I do know someone that used it to explore a number of legal issues they were involved in, and the answers gave a pretty good response, including letters that could be sent.

Careful though
There’s already cases of actual lawyers bringing AI **** to court as evidence (or witness impact testing at least) because as far as I can see AI feeds on AI…creating self perpetuating ‘facts’ (note the lower case!)

The use I have seen that makes sense is asking it to check your insurance documents for various potential claims.
No one has enough life to do that on your own so that makes sense to me

 

[dennisboothstash]

I use it almost daily to summarise information/create content for work.

We've saved the salary of a marketing assistant who moved on and I've not replaced.

ChatGBT isn't as nice to look at as she was, and probably won't flash it's arse at the office party (I'll miss that) but it's quicker, cheaper and only requires about the same amount of corrections/rewrites.

Hmm
And who, societally, is picking up the pieces of that job loss?
It should be the person who is making the most money from your company, but I haven’t seen any evidence to suggest that’s happening?

 

[DMD]

Careful though
There’s already cases of actual lawyers bringing AI **** to court as evidence (or witness impact testing at least) because as far as I can see AI feeds on AI…creating self perpetuating ‘facts’ (note the lower case!)

The use I have seen that makes sense is asking it to check your insurance documents for various potential claims.
No one has enough life to do that on your own so that makes sense to me

I'm not sure what you mean by that, but I'm always happy to learn from other peoples experience. Do you have an example?

Without boring you with the details, the person I know uses it for advice, and then checks through the relevant legislation and case law before proceeding. I've done similar several times with success, but without the help of A1. I'm not really sure what 'facts' the solicitors could use from that against them.

I'm all in favour of anything that minimises solicitors and increases the money available to the parties involved rather than lining their pockets.

 

[dennisboothstash]

I'm not sure what you mean by that, but I'm always happy to learn from other peoples experience. Do you have an example?

Without boring you with the details, the person I know uses it for advice, and then checks through the relevant legislation and case law before proceeding. I've done similar several times with success, but without the help of A1. I'm not really sure what 'facts' the solicitors could use from that against them.

I'm all in favour of anything that minimises solicitors and increases the money available to the parties involved rather than lining their pockets.

I see my late night eyes failed to see my mistype
I meant AI witness statements
https://www.facebook.com/share/v/1BTuvcHR25/?mibextid=wwXIfr
I’ve seen eminent lawyers describing this as a thin edge of a ****ty wedge, but I’m not qualified to say

Your example seems fine until they get lazy

I can just see (ok imagine) a time when ‘facts’ become accepted because some lie earlier has made it through an algorithm check and been accepted forever as fact by numerous bots.

It’s all dogshit…everyone should read the encyclopaedia Britannic

 

[PLT]

"Summarize the content on NOT 606 transfer thread (link) and who thinks what in general"



Obvious that it took a liking to you Stockholm Tiger

You must log in or register to see images

You must log in or register to see images

Haha, 'tends to express broader societal frustration'. It's not wrong.

Interesting also that it seems to be looking at other threads on the forum to make some of those judgements aside from the one you asked it to use.

 

[DMD]

I see my late night eyes failed to see my mistype
I meant AI witness statements
https://www.facebook.com/share/v/1BTuvcHR25/?mibextid=wwXIfr
I’ve seen eminent lawyers describing this as a thin edge of a ****ty wedge, but I’m not qualified to say

Your example seems fine until they get lazy

I can just see (ok imagine) a time when ‘facts’ become accepted because some lie earlier has made it through an algorithm check and been accepted forever as fact by numerous bots.

It’s all dogshit…everyone should read the encyclopaedia Britannic

Ah, I can totally understand that. I agree, using AI to write a 'witness' statement is against the fundamental purpose of that part of the process. I wouldn't want to be questioned in Court on a statement I effectively hadn't written, and that by definition is a fiction rather than contemporaneous fact.

 

[dazzar86]

Google's Veo 3 is interesting on the AI front.

You can type in commands for a story or script for your own movie trailer and it will output a video of decent quality within minutes. You can give descriptions of your characters and dialogue for them and even suggest an accent.

 

[tigerscanada]

I struggle to believe anyone's using it for legal letter or essays to any meaningful standard. Some of the responses it produces are just bollocks.

It's a step up from a search engine, if you know what you're looking for you can quickly separate the wheat from the chaff, but people using it blindly like that will soon get caught out.

The old adage "Garbage in, garbage out" applies most likely.

 

[x]

i'm finding gemini on my android phone really useful, apart from almost any movement of the phone making the app disappear. also sometimes a bit annoying trying to dictate a question somewhere noisy.

 

[Heimdallr]

Never used it. Give me a shout when it can cook a dinner, pacify an angry woman, feed a sheep and weld two bits of steel, whilst hanging off a rope in a storm, and it will serve a purpose in my life...

Joking aside, will it really be a new revolution? I appreciate it will in certain job sectors.. accountancy, marketing, work online etc but will the development really alter how we use our time. The industrial revolution effectively triggered women moving into paid work some decades later, together with many other factors, so hopefully if it is a revolution, we'll see all of us working less in a few decades time.. but it will be too late for me.

 

[THE EXCLUSIVE 10%]

I am just finalising setting up an A.I company which is designed to CREATE jobs and revenue. A.I isn’t perfect however it improves with every action. For most people the more detailed the prompt the more accurate the information given. The A.I the general public see and use is the first draft so to speak. What its true capabilities entail is mind blowing but won’t be released yet as we are not ready for it. As someone said, it’s a new revolution. Early adapters will be the ones that benefit most.

 

[eimaj]

The old adage "Garbage in, garbage out" applies most likely.

Most certainly - It's a glorified Search Engine today. Until it can reason, which I'm sure won't be long, it's as good as the content it's trained on.

I also hate how it thinks the sun shines out of the user's arse. "You're absolutely correct...", "That's a great idea...". When things aren't black and white, I wish it'd have a differing opinion every once in a while.

 

[originallambrettaman]

I had a play with ChatGPT a few months ago, I ran some text from my website through it and asked it to improve it and it did a decent job. Then I asked it some questions about my old business (being a fashion brand, there's a lot of information available online), just to see how accurate it was and it got most things completely wrong. It will obviously improve as time goes on, but it still feels quite primitive at the moment.

Far more impressive, is the @grok service on X, as a fact checking service it seems to be incredibly accurate.

There's hundreds of app's being advertised online at the moment, particularly in my trade. Things like uploading a flat picture of a garment to an app that puts it on a model of your choice, but I'm yet to come across a single one that actually works. I've no doubt that these things will come though.

 

[springtiger]

Is it ‘left’ or ‘far right’? LOL
Who has it been written by ?

 

[DMD]

Talking to Professors and academics fully engaged in exploring the possibilities from a variety of angles, and in my ignorance started them having an interesting debate with each other by me asking what the difference is between AI and machine learning.

As a summary, the consensus was that machine learning is far more advanced than AI and is what the majority of products actually are, but calling it AI is more 'sexy' so they don't tend to correct people if they use the wrong term as it impacts funding.

As a very broad rule of thumb, machine learning can only use what's already in its own data base to draw its conclusions from, whereas pure AI can operate without being pre-loaded with much background data.

A lot of the advancement in machine learning is associated with the advances in computer processing speeds meaning they can sort data so much faster than previously giving it the appearance of AI.

EDIT: I think we would maybe benefit from advances in Actual intelligence.

 

[dennisboothstash]

It’s not great in terms of energy use either

Ban it or heavily tax it

 

[DMD]

It’s not great in terms of energy use either

Ban it or heavily tax it

Going down the tinfoil hat route, there is a school of thought that the servers for AI, which as you say are massive drains on energy, water and other resources, are supported by governments as they will end up being vast repositories of data on people, with alarming capabilities on how to draw 'conclusions' from that data.

Several years ago, I saw a series of presentations by experts in data analysis, and it was both impressive and alarming how much information they could draw from what seemed to be minimal and unconnected pieces of information, and that's without even applying AI. Their conclusions were frighteningly accurate.

Their examples looked like Sherlock Holmes on steroids.

I've also seen computer people that think they are data analysts because the crunch numbers, as well as public health 'experts' that didn't understand data management but set national policies, and their conclusions were woeful.

For me, there is a big risk in who is deciding which projects move forward the furthest and how the data is used and managed.

 

[Stockholm Tiger]

Interesting you say we have more leisure time as a population as a whole.

I don’t have any facts to back this up, but my gut feeling is that’s possibly not true over the last 40/50 years. Previously, many families survived on one income. That’s not the case for lots these days with two incomes being the norm - squeezing leisure time to weekends when other chores will also need to be done.

I know if you look back further past that time you’ll see many men doing 6/7 day weeks down the pit etc but I think if you average out across genders etc the stats might not be as favourable as technological advancements may have initially suggested since the 70s.

Im more than happy to be proved wrong here as I say this is more my gut feeling than anything rooted in fact.

I don't have an academic paper for the last 50 years to be honest Drew so I can't answer directly with any honesty.

However, I do know that between the mid-60s and the mid-2000s people on average gained around 6 hours additional leisure time.

I could believe this trend started to reverse during the second half of the 2000s and 2010s but I haven't seen a paper confirming this.

However, since the pandemic there has been a massive switch in the acceptance of remote working and/or hybrid working with people in the office 1 or 2 days per week. I suspect the removal of all of those commuting hours may give on average increased "leisure" time, which is a result of the 4th IR delivering robust remote working tools such as Teams. (Of course this benefit is lumpy not everyone can WFH and commute time vary drastically. TWT)

It's an interesting topic.