Chat GPT & Generative AI tools

[Trader724]

Web browsing and Plugins are now rolling out in beta (May 12)
If you are a ChatGPT Plus user, enjoy early access to experimental new features, which may change during development. We’ll be making these features accessible via a new beta panel in your settings, which is rolling out to all Plus users over the course of the next week.




Once the beta panel rolls out to you, you’ll be able to try two new features:

Web browsing: Try a new version of ChatGPT that knows when and how to browse the internet to answer questions about recent topics and events.

Plugins: Try a new version of ChatGPT that knows when and how to use third-party plugins that you enable.



To use third-party plugins, follow these instructions:

Navigate to https://chat.openai.com/

Select “Plugins” from the model switcher

In the “Plugins” dropdown, click “Plugin Store” to install and enable new plugins



To enable beta features:

Click on 'Profile & Settings'

Select 'Beta features'

Toggle on the features you’d like to try.

Nice

[ShaunWhite]

looking to build my own ai for a non-trading related project. Any thoughts on the best place to start with that?

Scikit or Power BI.?

[arbitrage16]

I guess it depends on what you are trying to do.

There is a lot of hype out there, so be careful. A lot of stuff positioned as AI simply isn't and there seems to be a rush to push ChatGPT API products as though they are unique rather than a subset of LLM like Chat GPT.

To feed it with books and novels and other text documents in order that I can ask it for examples of particular aspects from a wide variety of texts.

ChatGPT is currently horrible at this, the phenomena of AI hallucinations heavily in evidence.

[sionascaig]

Looks like FBref have stopped making their data easy to scrap (removed the convert to csv, excel, etc options)...

When I looked a bit more into it they have a notice saying due to the rise in AI based requests have taken a couple of actions:

1. Any AI application is not allowed to use data as per terms
2. Bespoke data request now have a min charge of £1000 per request...

Its more complicated than that as a number of data owners involved but that is the gist..

[sionascaig]

Can see a lot more folk trying to protect their data from AI algos - a lawyers dream about to come true.

[Euler]

On a bing summary I asked for recently, I was quoted as a source. It does make you think about how these LLM have the potential to be giant great thieves on a monumental scale.

[ShaunWhite]

Scraping data isn't new, it's only now AI has been made accessible and anthropomorphised people have started noticing it.

It's all good imo.
https://www.bbc.co.uk/news/health-65709834.amp

[wearthefoxhat]

Governments/Corporations always like to control new innovations, AI is no different.

IMO, it's the best learning tool to help level the playing field in all walks of life.

[megarain]

I have the $20 a month Beta addon etc.

It wasnt working great, but this morning a Bing plugin came available.

To test, I put this question in :

It sorted a result in under a minute. It is pretty impressive (thou I suppose how useful the data is, is debateable).

I dont know if the results are accurate.

Can you rank the top 10 racecourses in the UK where pre-race favourites win the most, since 1 Jan 2020

ChatGPT
Finished browsing
The top 10 racecourses in the UK where pre-race favourites win the most, since 1 Jan 2020, are:

Carlisle (48% non-handicap, 32% handicap)
Thirsk (51% non-handicap, 29% handicap)
Windsor (49% non-handicap, 33% handicap)
Yarmouth (48% non-handicap, 36% handicap)
Catterick (58% non-handicap, data for handicap races not provided)
Hamilton (50% non-handicap, 28% handicap)
Brighton (44% non-handicap, 31% handicap)
Bath (43% non-handicap, 34% handicap)
Lingfield (44% non-handicap, 33% handicap)
Sandown (44% non-handicap, 41% handicap)

Please note that these percentages reflect the total win rate for favourites across both handicap and non-handicap races. Also, the data for Catterick does not provide a win rate for handicap races, so the overall win rate may be different if this data were available​1​.

[wearthefoxhat]

I have the $20 a month Beta addon etc.

It wasnt working great, but this morning a Bing plugin came available.

To test, I put this question in :

It sorted a result in under a minute. It is pretty impressive (thou I suppose how useful the data is, is debateable).

I dont know if the results are accurate.

Can you rank the top 10 racecourses in the UK where pre-race favourites win the most, since 1 Jan 2020

ChatGPT
Finished browsing
The top 10 racecourses in the UK where pre-race favourites win the most, since 1 Jan 2020, are:

Carlisle (48% non-handicap, 32% handicap)
Thirsk (51% non-handicap, 29% handicap)
Windsor (49% non-handicap, 33% handicap)
Yarmouth (48% non-handicap, 36% handicap)
Catterick (58% non-handicap, data for handicap races not provided)
Hamilton (50% non-handicap, 28% handicap)
Brighton (44% non-handicap, 31% handicap)
Bath (43% non-handicap, 34% handicap)
Lingfield (44% non-handicap, 33% handicap)
Sandown (44% non-handicap, 41% handicap)

Please note that these percentages reflect the total win rate for favourites across both handicap and non-handicap races. Also, the data for Catterick does not provide a win rate for handicap races, so the overall win rate may be different if this data were available​1​.

Note too sure TBH...


By highest SR%

HighestSR%.png

By most profitable

Flat UK.png

[megarain]

Lols - thx for checking.

It will clearly evolve and maybe my prompts need to be refined - but can see online search changing a lot.

[firlandsfarm]

This thread seems to have dried up ... there are lots of AI bots on the Internet now specialising in various subjects. Just wondered if anyone has found one claiming to be targeted towards our topics and if they think it's any good?

I recently came across this statement ... "AI will look for patterns that emerge in large datasets and is being already used by large betting companies ..." (https://thedailytipster.co.uk/chatgpt-a ... s-betting/). Does this mean value will become more and more elusive? The bots mentioned in that article seem to be simply AI based tipsters.

I have no experience of using ChatGPT or a similar 'front line' bot but are people putting some form of front end door to the bot that asks the bot to publish it's findings perhaps daily?

[decomez6]

AI will look for patterns that emerge in large datasets and is being already used by large betting companies .

a charity shop of AI trading bots .. OR some sorts Scrap Yard for used betting patterns

=garbage in /garbage out ..... A I is only useful if you have an accurate Database. & NO body (even RED CROSS relief) will share a winning formula .

imagine two winning BOTS against each other. ....I will use my human intellect , let them nihilate each other as i pick up thier hard work , RUN HOME clean and DRY

[Euler]

This thread seems to have dried up ... there are lots of AI bots on the Internet now specialising in various subjects. Just wondered if anyone has found one claiming to be targeted towards our topics and if they think it's any good?

I recently came across this statement ... "AI will look for patterns that emerge in large datasets and is being already used by large betting companies ..." (https://thedailytipster.co.uk/chatgpt-a ... s-betting/). Does this mean value will become more and more elusive? The bots mentioned in that article seem to be simply AI based tipsters.

I have no experience of using ChatGPT or a similar 'front line' bot but are people putting some form of front end door to the bot that asks the bot to publish it's findings perhaps daily?

ChatGPT is a large language model, it's not a statistical model. Lots of people jumping on the bandwagon to promote 'AI' services just shows how out of touch they are IMHO.

[greenmark]

AI will look for patterns that emerge in large datasets and is being already used by large betting companies .

a charity shop of AI trading bots .. OR some sorts Scrap Yard for used betting patterns

=garbage in /garbage out ..... A I is only useful if you have an accurate Database. & NO body (even RED CROSS relief) will share a winning formula .

imagine two winning BOTS against each other. ....I will use my human intellect , let them nihilate each other as i pick up thier hard work , RUN HOME clean and DRY

I don't really understand AI but I've read that some AI engines are "educated" by humans. The AI engine answers questions and humans correct the answers (at low pay). Not exactly Asimov's "I robot" is it?