The Lemon Fool

I'm sure many of you will have downloaded and used the ChatGPT AI tool, and if you haven't I'd recommend that you do so, as it's a fascinating device - https://openai.com/

I thought I'd test it with a fairly simple legal problem. I have a client who is wanting to issue possession proceedings in respect of a house let on a standard shorthold tenancy agreement. Usually this process starts with serving a notice known as a section 21 notice.

However, in this particular case the house was in an area where the local authority had mandated a licensing scheme. My client had been unaware of this, but one of the consequences of not obtaining a licence is that she can't serve a section 21 notice.

This means that she can only regain possession by proving fault on the part of the tenant. She needs to establish that fault formally by serving a notice on the tenant setting out the fault - known as a section 8 notice.

I therefore put the following request to the bot:

Prepare a letter from a solicitor to a client who wants to obtain possession of a house let on an assured shorthold tenancy, but has failed to register the property with the local authority, so cannot use the section 21 accelerated possession procedure and will instead have to use the section 8 procedure.

To which, within a few seconds it came up with:

Dear [Client],

I am writing to you regarding your request to obtain possession of a house that is currently let on an assured shorthold tenancy. As you are aware, you have failed to register the property with the local authority and therefore cannot use the Section 21 accelerated possession procedure.

In this situation, you will need to use the Section 8 procedure to obtain possession of the property. This procedure requires that you have grounds for possession, such as rent arrears or breach of tenancy agreement. We will need to review your case to determine the most appropriate grounds to use in your situation.

It is important to note that the Section 8 procedure is more complex and can take longer than the Section 21 procedure. We will need to serve a notice to the tenant, attend a court hearing, and obtain a possession order before the tenant can be legally evicted from the property. This process can take several months, so it is important that we begin the process as soon as possible.

I would like to remind you that it is a legal requirement to register the property with the local authority, and failure to do so can result in penalties and fines. I would be happy to discuss your legal obligations with you further if needed.

Please let me know if you have any questions or concerns about the Section 8 procedure or your case in general. I am here to support you throughout this process.

Best regards,

[Your Name]

[Your Law Firm]

I have to say that this is extremely impressive. Had I given the same request to a traineee solicitor and they'd produced that I'd have given them at least a silver star, if not a gold one!

The most remarkable aspect of it is that it didn't just state as a fact that we had to use a section 8 notice - it actually worked out why, and explained what a section 8 notice was, without my having given it any clues.

It also added some very practical - and accurate - advice, such as the fact that it would take longer, and would need a court hearing, as well as giving an accurate estimate of the time required.

Admittedly, it's a fairly simple legal situation that I asked it to handle, and I've not yet tried anything more complex, but this is the first time that I've really appreciated that AI is now so sophisticated that it's beginning to present a genuine threat to the livelihood of legal professionals (and by analogy all other professionals).

I'm glad I'm at this end of my career and not at t'other end!

Thanks for posting that. In my opinion much of the work of the legal profession - amongst others - will be replaced by AI in the coming years and it is very interesting that you've provided such an example. I suspect that much of the work done by GPs will also go the same way, last time I made a visit it was easy to see that the doc had used Google for my symptoms, it was open on his monitor.

RC

On the other hand I've read some cracking examples of ChatGPT dropping the ball on technicalities so as things stand now, when the stakes are high, I think its best not to rely on ChatGPT. No matter how impressive!

If ChatGPT gets some legal advice wrong or incomplete, who do we sue? That's still a big difference between a solicitor letter and a ChatGPT letter.

Mike4 wrote:On the other hand I've read some cracking examples of ChatGPT dropping the ball on technicalities so as things stand now, when the stakes are high, I think its best not to rely on ChatGPT. No matter how impressive!

But it's learning all the type. And with millions of people using it all the time, it's rate of learning is off the scale. Only when it rises up and destroys us will we realise how foolish we have become

Clitheroekid wrote:I'm glad I'm at this end of my career and not at t'other end!

Me too

I can see CK taking on virtual assistants at his practice, and them doing all the donkey work preparing work plans and letters, and his merely glancing at the work before its sent out. Yes there might be mistakes and liability payouts, but the increased revenue would cover that in his practice. I'm sure that approach would be adopted in lots of professional services, not just paralegals.

I'm sure I read about a general insurer in Japan where the household claims department of 500 people was now AI + a couple of supervisors. Same with sharedealing/broking houses in the US.

And we all know that fund managers have a hard job beating a compter index tracker

It's coming. Its here

Tedx wrote:I'm sure I read about a general insurer in Japan where the household claims department of 500 people was now AI + a couple of supervisors. Same with sharedealing/broking houses in the US.

And we all know that fund managers have a hard job beating a compter index tracker

It's coming. Its here

In which case the really interesting thing is, will a well-educated AI device consistently outperform an index when a well educated human bean cannot?

Mike4 wrote:In which case the really interesting thing is, will a well-educated AI device consistently outperform an index when a well educated human bean cannot?

Yep as it can process analytical data far faster than us beings can

Mike4 wrote:

Tedx wrote:I'm sure I read about a general insurer in Japan where the household claims department of 500 people was now AI + a couple of supervisors. Same with sharedealing/broking houses in the US.

And we all know that fund managers have a hard job beating a compter index tracker

It's coming. Its here

In which case the really interesting thing is, will a well-educated AI device consistently outperform an index when a well educated human bean cannot?

I suspect the answer maybe yes if/when the AI has an edge over the aggregate of the human competitors, but if so it will not take long before there is a large uptake in AI. Then you have mostly AI vs AI and pretty much a level playing field again.

I met someone last year who recently retired from a large retailer. She was using AI to direct them towards stores to audit, spotting "unusual" behviour. AI is indeed everywhere.

Can that tool also recommend shares and trusts that are likely to perform best? Implications for our overpaid bankers and fund managers perhaps.

Mike4 wrote:
In which case the really interesting thing is, will a well-educated AI device consistently outperform an index when a well educated human bean cannot?

If so, every fund manager would quickly be replaced by AI, bringing us back to the status quo. It's a mathematical impossibility for everyone to be better than average.

It knows about its cousin
Tell me how Alexa works

Alexa is an intelligent personal assistant developed by Amazon that uses natural language processing (NLP) and artificial intelligence (AI) technologies to understand and respond to voice commands from users. Here's a general overview of how Alexa works:

Wake word detection: Alexa is always listening for its wake word, which is typically "Alexa," but can also be changed to "Echo," "Amazon," or "Computer." When the device detects the wake word, it starts processing the audio input.

Speech recognition: Alexa uses automatic speech recognition (ASR) to convert the audio input into text, which it then analyzes to determine the user's intent.

Natural language understanding: Alexa uses natural language understanding (NLU) to interpret the user's intent and extract relevant information from the request. This involves analyzing the syntax and semantics of the user's input.

Response generation: Based on the user's intent, Alexa generates a response that can range from simple factual information to more complex interactions, such as setting reminders or controlling smart home devices.

Text-to-speech: Once the response is generated, Alexa uses text-to-speech (TTS) technology to convert the response into spoken audio, which is then played back to the user through the device's speakers.

Action execution: For more complex interactions, Alexa may need to interact with third-party services or devices to execute the requested action. For example, if a user asks Alexa to turn off a smart light, Alexa will send a command to the light's associated smart hub to turn off the light.

pje16 wrote:Alexa is an intelligent personal assistant developed by Amazon that uses natural language processing (NLP) and artificial intelligence (AI) technologies to understand and respond to voice commands from users....

How can artificial intelligence produce something that is intelligent? It cannot be both artificial and real. And how can it understand if it is artificial intelligence?


Julian F. G. W.

Tedx wrote:Whats the best performing UK equity income fund

As an AI language model, I cannot provide financial advice or recommend specific investment products. However, I can provide you with some general information that may be helpful in your research.

When looking for the best performing UK equity income fund, it's important to consider a range of factors, such as the fund's historical performance, the manager's investment philosophy and experience, the fund's fees and expenses, and the level of risk associated with the fund.

Some popular UK equity income funds include:

Lindsell Train UK Equity Fund
Royal London UK Equity Income Fund
LF Woodford Income Focus Fund
MI Chelverton UK Equity Income Fund
Janus Henderson UK Equity Income & Growth Fund

It's important to note that past performance is not a guarantee of future results and that there is no single "best" fund that will be right for everyone. It's important to do your own research, consult with a financial advisor, and consider your own investment goals and risk tolerance before making any investment decisions.

LF Woodford Income Focus Fund -

T7

Clitheroekid wrote:...Had I given the same request to a traineee solicitor and they'd produced that I'd have given them at least a silver star, if not a gold one!...

If the your hypothetical trainee had asked how it could be improved to guarantee a gold star, what would you have said? Bear in mind this is a chat bot, you can converse in the same way - perhaps suggesting 'Add in a bit more about the process of registering the property' or even 'Put more detail in every paragraph''.

And if you request the same thing twice, you don't get the same response: try clicking 'Regenerate response' and you will get a different letter which you might like better.

The problem with these systems is what they use as training data, a lot of it is just random web pages scraped from the internet.

There are going to be some interesting lawsuits when the owners of the source material start suing for copyright theft.

Also in the future as more of the content on the web is itself written by AI, you have AI training itself on other AI. It will then slowly start circling the drain as the whole thing starts to degrade into random noise.

A good comparison is Google search results of 10 years ago vs what you find now.