I've been experimenting with ChatGPT since last month and it is truly amazing. But you are right that it can and will make errors. I asked it who Michael Owen is and it gave a good answer talking about his career with Liverpool, Madrid and England but right at the end said "he was a member of the 1966 World Cup winning team". I then asked for his date of birth, which it gave correctly, so that was odd. As you say, it is a language generation tool so it doesn't seem to have the ability to check basic logic like this in its answers. Having said that my mind has been blown by 99% of the conversations I've had with it.
chatgpt (at the moment, in its infancy) is the next evolution in search engines. Microsoft has recently invested around 1.5B in Chatgpt, and Google is a bit apprehensive, as it could quite easily and quickly replace them in the global search marketplace. interesting times ahead:)
So the biggest trouble is that this algorithm, as amazing as it is, is merely a prediction based model, it literally puts together words based on the probability of the same words appearing one after each other in its data set. It basically does not understand what you're asking, or the replies it gives, it links bits and pieces together to provide you an answer that is "believable". This is why sometimes it can give you extremely weird or even contradicting results. But you're right in saying that it is going to change a lot of different fields, for the better or the worse. Now... why I do think things can go extremely bad? Imagine people start using this algorithm to write content, lots of content for cheap, without fact checking it. I can already see it happening in blog posts. Now imagine that this content is then swiped and used as an update in the model itself, to provide new content. The more the cycle is repeated, the more you will get plausible answers that are completely false, and no one checking them. (like a summary of Rich dad poor dad that is indeed the summary of Cinderella)
I use ChatGPT, but I would never claim its output as my own. I put a comment at the top saying "Author: ChatGPT" and I also list the prompts that generated the output. If there's a problem with the output, the people that come to fix the problem after I have gone will need to know where it came from. Tend to agree that people will try to pass the output off as their own work and the AI output will end up being incorporated into the AI input. MS have already just (lazily, in my opinion) simply connected their AI directly to the Internet (what could possibly go wrong?).
I've tried CharGPT with multiple tasks including programming in languages I understand and solving questions. It is very good at giving you an answer but often gives an incorrect one.
It takes away the human emotions from the investing process so it can be a great tool, but overall I am worried: Humanity strive when can overcome problems and evolve but this is a tool that makes all of us way more lazy... our brain is our best asset and we should keep it sharp and trained
I’m an machine learning architect. This is a good example of the fact that ChatGPT just looks for correlations but can’t reason. It justified its fund recommendations like that is because it worked out that when the question ‘why did you choose these funds’ was asked before, the answer usually involves having good performance. This must happen many times in its training data. But no-one ever justifies a poor performing fund by saying it has good performance. That data never appears. But it has no notion of what performance actually is, or why the performance of the fund is, it just blindly finds correlations, resulting in logical errors
like your content James - literally just came across you. Question: would you be willing to be interviewed? I'm drilling down to women and their financial facts of life. Would be interested in your take on what they could/should do/ look out for etc. Let me know and we can exchange emails?
The cynic in me says....if the AI is 90% good, then an accountant will say that's good enough to replace the expensive human and the misses will just be collateral damage.
If you think accountants have the power to make such decisions unilaterally, or believe that they think this way, then you don't really know much about what they do...
I am a Risk Manager, I think the biggest changes might well come in corporate Business models. I already pointed this out to one of my Insurance clients who decided they would ignore it. Doing so in my opinion means their business model will be dead in the next 20 years. As a board you are forward facing when setting strategy and if you do not listen to people like me you may have no future business. AI has massive upside and downside, but you need to be more open minded because as you say there will be monumental changes to come and they will increase in speed.
Great video. Could you link me to that video you did where you pretended to be on the TV and helped your identical twins to invest? It’s very hard to find
Hi @JamesShack the ChatGPT knowledgebase cut-off date is end of 2021 - so anything it says about the stock market returns are only up to that point. This can be grossly inaccurate as you know. It should really qualify it's responses in light of that. Even if you specify in your query "in the last 12 months"... it will use the 12 months from 2020 to 2021. e.g. ETF/Index fund returns. Q: Who is the current UK prime minister? A: As of my knowledge cut-off date, 2021, the current UK prime minister is Boris Johnson. He has been in office since July 2019, succeeding Theresa May. He is a member of the Conservative Party and leads the UK government.
Financial peace isn't the acquisition of stuff. It's learning to live on less than you make, so you can give money back and have money to invest. You can't win until you do this. --Dave Ramsey;
I think if this GPT was allowed to be updated to the web daily it would be amazing. The first thing i can think of is showing me trends of lottery numbers for pick 3 and 4. And going from there.
Thanks James, for the video, like you I have found errors when asking questions about narrow topics or events. It provides wrong answers that someone with no knowledge or little knowledge about the topic will believe.
James, apologies but I am struggling to find a video that I watched previously today, it related to death benefits whereby you talked through an example whereby it cost the individual £400k in inheritance tax charges. Brilliant video but for some reason I can't find it again for some reason. Can you advise if it is still accessible/ viewable please?/
@@JamesShack Much appreciated James, in the process of writing a will and toying with the idea of powers of attorneys, tenancy in commom etc thats all. It was very pertinent👍
I see limitations though. If its information source is the Internet, it can be manipulated by sites that it indexes if they show fake or biased information. How long before those indexes are manipulated? I bet they already are, just like Google decides what you see in your search results, the chatbots will be leading us to what it wants us to know, as decided by some unknown entity.
I hope AI replaces many law jobs, which in the area of Conveyancing and Probate, are incredibly expensive and yet just follow a series of rules and regulations that I'm sure AI could be far more efficient at.....£200/hour for a paralegal could be replaced! Just been through Probate, my father's house sat empty for 8 months, imagine what a dent could be made in the housing crisis if this time could be halved!
I can imagine how this could be used to actually manipulate the market, rather than to give advice. Similar in some ways to Google or Facebook removing sites from searches where its politics don't fit.
But there will be no savings under the New Central Bank Digital Currencies. Perhaps in the short term, to get us onto the new system, but not in the longer future.
Brilliant simplistic video as always James. Your skill isn’t just your in-depth knowledge, it’s the ability to break things down in laymen’s terms. However ChatGPT is flawed in that it represents its creators left wing, woke views and it is still about its creators biases rather than being objective. The potential is limitless once the biases are removed
I like this vid. I agree with you it is a great accompanying tool. At the moment it is not accurate enough to replace the Human. I am to Human for chatGPT.
Everyone using ChatGPT will produce the same results. The scope for massive shifts in stock prices increases, and more frequently. Probably not a good thing. It reduces the individuality of your existence. Put the work in, be original, and you'll stand out.
chat gpt adds more bias than just google/duck duck go searching. I have been trying to trick it by asking hot-topic subjects and seeing how well it defines logic and from what angle/motivation. Questions that would straddle political/financial divide etc. Like a lot of ai you can quickly see patterning in the logic. Human restricted Patterning. I was trying to interrogate it on stock market technical analysis and am impressed how easily it can simplify technical terms to a readily understandable language but disappointed at how nuanced it would allow itself to get by restricted speech and clauses. This ultimately makes ai vulnerable to bias and manipulation. I will definitely use ai in future analysis as a first port of call on any search though.
I'm not concerned at all. Robo advice has been around for years and that's more geared towards people with simple situations. They're not my typical client anyway, so I don't feel the threat.
ChatGPT does not make decisions and doesn't understand the concepts it writes. In it's current form it can't change the financial industry, all the info about index funds and (under) performance of hedge funds is available for years to everyone.... yet people still buy APPL because it's a "great company".
it's also OFFLINE since 2019 or so, so all info it gives is 3 yrs out of date. Fine for history or anything not time/date sensitive but forget any info on best funds for today
Only if you have access to all the underlying data on which to base answers - which is not possible. Much of the source data is controlled by existing companies who are working on the same AI/ML solutions already.
Be very very careful with chatGPT. It absolutely has zero concept of British tax regulations. Do not ask it for advice and help with your self assessment or anything else along those lines. It omits things and when you point it out it basically says "Oh yeah, so the answer is probably that then."
If I could flood the internet with comments saying Where's Wally is the best book on wealth creation, that will be the book ChatGPT will recommend when asked. It is extremely sophisticated technology, but the ability to be gamed in that way suggests to me that there is a long way to go.
@@JamesShack It will improve in simulating intelligence. It will never be intelligent. Simulating intelligence is all any algorithm in the much hyped and misnamed field of 'Artificial Intelligence' can ever do.
@@JamesShack haha. Good question. I guess that I'd look for it to have the ability to create a novel perspective - and for it to be opinionated. Ultimately, your question becomes a rabbit-hole. Like asking "what is consciousness?" Love the channel by the way.
@@JamesShack Science really does not understand intelligence. We know lots about the physics, biology and chemistry of single neurons but we know very little about how neurons act together to become conscious. It still seems almost like a miracle.
Hahahahahahaha… so you trust an AI solution run by a faceless corporate… “Tell me who I should invest my hard earned life savings in?…”. Yeah, I’ll trust that…!!!
Financial service fees have been coming down as technology and competition have proliferated. This technology will reduced the costs of providing advice enormously. So as long as the market is still competitive costs to consumers will reduce dramatically too.
Interesting question, I work healthcare performance and analytics. I feel like it would certainly help with the heavy lifting enabling more time in fine tuning. Funding that will be another matter 🫠, but that is another discussion not for this channel
I think businesses like OpenAI will develop into a Model as a Service business, leaving other companies to work out how to implement it for different use cases.
Asking ChatGPT for the best funds is like typing it into google and assuming the top 5 are the best. ChatGPT doesn't understand what makes a good fund, it just knows how people write about funds and makes some text up based on that. You be better building an AI stock model like I have, than using a language model. Predicting what you want to hear based on what other people have typed online isn't a good way to model finances lol
ChatGPT is just the text equivalent of a pretty new GUI. There’s no intelligence to the responses it provides - it’s just presenting the information in a more human-readable, natural language format. There are definitely many developments in Machine Learning / AI that are focused on predicting good investments - but these are nothing to do with ChatGPT. ChatGPT could be the user interface that explains these investments to you in a way that a human would have done - but it’s still not the bit of code that’s making those financial decisions - it’s working out how to explain them to you in a way that feels like it’s human.
![The BIG Financial Risk That Everyone is Underestimating [Must Watch]](/img/n.gif)
How do you think A.I. like ChatGPT will change your industry?
I've been experimenting with ChatGPT since last month and it is truly amazing. But you are right that it can and will make errors. I asked it who Michael Owen is and it gave a good answer talking about his career with Liverpool, Madrid and England but right at the end said "he was a member of the 1966 World Cup winning team". I then asked for his date of birth, which it gave correctly, so that was odd. As you say, it is a language generation tool so it doesn't seem to have the ability to check basic logic like this in its answers. Having said that my mind has been blown by 99% of the conversations I've had with it.
What I think is that at this rate of AI development my job will be redundant pretty soon!!! AI robots don't need to be paid a salary.
It'll change almost every industry, this is a new Industrial Revolution. Hard to overstate.
It’s wonderful, but it does lie and made up some research papers when I used it for work
It'lll be the death of google one day.
Who needs AI when there's James?
Greg secker maybe?
@@stevie3452 haha good old Greg Secker.
My current summation of ChatGPT is that it’s a great tool for providing ideas, but it does require a human with expertise to fully process the ideas.
chatgpt (at the moment, in its infancy) is the next evolution in search engines. Microsoft has recently invested around 1.5B in Chatgpt, and Google is a bit apprehensive, as it could quite easily and quickly replace them in the global search marketplace.
interesting times ahead:)
So the biggest trouble is that this algorithm, as amazing as it is, is merely a prediction based model, it literally puts together words based on the probability of the same words appearing one after each other in its data set. It basically does not understand what you're asking, or the replies it gives, it links bits and pieces together to provide you an answer that is "believable". This is why sometimes it can give you extremely weird or even contradicting results. But you're right in saying that it is going to change a lot of different fields, for the better or the worse. Now... why I do think things can go extremely bad? Imagine people start using this algorithm to write content, lots of content for cheap, without fact checking it. I can already see it happening in blog posts. Now imagine that this content is then swiped and used as an update in the model itself, to provide new content. The more the cycle is repeated, the more you will get plausible answers that are completely false, and no one checking them. (like a summary of Rich dad poor dad that is indeed the summary of Cinderella)
I use ChatGPT, but I would never claim its output as my own. I put a comment at the top saying "Author: ChatGPT" and I also list the prompts that generated the output. If there's a problem with the output, the people that come to fix the problem after I have gone will need to know where it came from. Tend to agree that people will try to pass the output off as their own work and the AI output will end up being incorporated into the AI input. MS have already just (lazily, in my opinion) simply connected their AI directly to the Internet (what could possibly go wrong?).
I've tried CharGPT with multiple tasks including programming in languages I understand and solving questions. It is very good at giving you an answer but often gives an incorrect one.
Think of it as a Beta version, it will improve very fast, Chat GPT-4 is around the corner, orders of magnitude bigger model.
It takes away the human emotions from the investing process so it can be a great tool, but overall I am worried: Humanity strive when can overcome problems and evolve but this is a tool that makes all of us way more lazy... our brain is our best asset and we should keep it sharp and trained
I’m an machine learning architect. This is a good example of the fact that ChatGPT just looks for correlations but can’t reason. It justified its fund recommendations like that is because it worked out that when the question ‘why did you choose these funds’ was asked before, the answer usually involves having good performance. This must happen many times in its training data. But no-one ever justifies a poor performing fund by saying it has good performance. That data never appears. But it has no notion of what performance actually is, or why the performance of the fund is, it just blindly finds correlations, resulting in logical errors
Great video James. Food for thought.
like your content James - literally just came across you. Question: would you be willing to be interviewed? I'm drilling down to women and their financial facts of life. Would be interested in your take on what they could/should do/ look out for etc. Let me know and we can exchange emails?
The cynic in me says....if the AI is 90% good, then an accountant will say that's good enough to replace the expensive human and the misses will just be collateral damage.
If you think accountants have the power to make such decisions unilaterally, or believe that they think this way, then you don't really know much about what they do...
I am a Risk Manager, I think the biggest changes might well come in corporate Business models.
I already pointed this out to one of my Insurance clients who decided they would ignore it. Doing so in my opinion means their business model will be dead in the next 20 years.
As a board you are forward facing when setting strategy and if you do not listen to people like me you may have no future business.
AI has massive upside and downside, but you need to be more open minded because as you say there will be monumental changes to come and they will increase in speed.
Great video. Could you link me to that video you did where you pretended to be on the TV and helped your identical twins to invest? It’s very hard to find
ua-cam.com/video/oCBQekcu8jw/v-deo.html
This is going to be a game changer for the working class. Imagine asking AI to invest £20 per month.
Hi @JamesShack the ChatGPT knowledgebase cut-off date is end of 2021 - so anything it says about the stock market returns are only up to that point. This can be grossly inaccurate as you know. It should really qualify it's responses in light of that. Even if you specify in your query "in the last 12 months"... it will use the 12 months from 2020 to 2021. e.g. ETF/Index fund returns.
Q: Who is the current UK prime minister?
A: As of my knowledge cut-off date, 2021, the current UK prime minister is Boris Johnson. He has been in office since July 2019, succeeding Theresa May. He is a member of the Conservative Party and leads the UK government.
James. That is an awesome video. Well done I am going to dive in.
Excellent. Thank you
Financial peace isn't the acquisition of stuff. It's learning to live on less than you make, so you can give money back and have money to invest. You can't win until you do this. --Dave Ramsey;
you have to check everything, including research paper summary.
I think if this GPT was allowed to be updated to the web daily it would be amazing. The first thing i can think of is showing me trends of lottery numbers for pick 3 and 4. And going from there.
Thanks James, for the video, like you I have found errors when asking questions about narrow topics or events. It provides wrong answers that someone with no knowledge or little knowledge about the topic will believe.
James, apologies but I am struggling to find a video that I watched previously today, it related to death benefits whereby you talked through an example whereby it cost the individual £400k in inheritance tax charges. Brilliant video but for some reason I can't find it again for some reason. Can you advise if it is still accessible/ viewable please?/
Hi Darren, I noticed an error in the video so I’ve had to take it down. I’ll get it fixed and re-upload tomorrow 👍🏻
@@JamesShack Much appreciated James, in the process of writing a will and toying with the idea of powers of attorneys, tenancy in commom etc thats all. It was very pertinent👍
I see limitations though. If its information source is the Internet, it can be manipulated by sites that it indexes if they show fake or biased information. How long before those indexes are manipulated? I bet they already are, just like Google decides what you see in your search results, the chatbots will be leading us to what it wants us to know, as decided by some unknown entity.
Sourcing is going to be hard. And most questions don’t have a definitive right answer.
I hope AI replaces many law jobs, which in the area of Conveyancing and Probate, are incredibly expensive and yet just follow a series of rules and regulations that I'm sure AI could be far more efficient at.....£200/hour for a paralegal could be replaced! Just been through Probate, my father's house sat empty for 8 months, imagine what a dent could be made in the housing crisis if this time could be halved!
I can imagine how this could be used to actually manipulate the market, rather than to give advice. Similar in some ways to Google or Facebook removing sites from searches where its politics don't fit.
I’ve got to admit, I’m one of many that are only just hearing about this AI stuff at the end of May 23. It’s crazy for a 56 yr old.
When asking a question of ChatGPT, how do you know it does not know who you (JA) are. So it will tailor its answer.
But there will be no savings under the New Central Bank Digital Currencies. Perhaps in the short term, to get us onto the new system, but not in the longer future.
Brilliant simplistic video as always James. Your skill isn’t just your in-depth knowledge, it’s the ability to break things down in laymen’s terms. However ChatGPT is flawed in that it represents its creators left wing, woke views and it is still about its creators biases rather than being objective. The potential is limitless once the biases are removed
I like this vid. I agree with you it is a great accompanying tool. At the moment it is not accurate enough to replace the Human. I am to Human for chatGPT.
Skynet is here 😀
Everyone using ChatGPT will produce the same results. The scope for massive shifts in stock prices increases, and more frequently. Probably not a good thing.
It reduces the individuality of your existence. Put the work in, be original, and you'll stand out.
chat gpt adds more bias than just google/duck duck go searching. I have been trying to trick it by asking hot-topic subjects and seeing how well it defines logic and from what angle/motivation. Questions that would straddle political/financial divide etc. Like a lot of ai you can quickly see patterning in the logic. Human restricted Patterning. I was trying to interrogate it on stock market technical analysis and am impressed how easily it can simplify technical terms to a readily understandable language but disappointed at how nuanced it would allow itself to get by restricted speech and clauses. This ultimately makes ai vulnerable to bias and manipulation. I will definitely use ai in future analysis as a first port of call on any search though.
Hope you don't mind but I'm going to forward this video on my Facebook. I'm glad you tackled this one. AI is going to change many professions.
Go for it.
Planning is easy
I'm not concerned at all. Robo advice has been around for years and that's more geared towards people with simple situations.
They're not my typical client anyway, so I don't feel the threat.
ChatGPT does not make decisions and doesn't understand the concepts it writes. In it's current form it can't change the financial industry, all the info about index funds and (under) performance of hedge funds is available for years to everyone.... yet people still buy APPL because it's a "great company".
it's also OFFLINE since 2019 or so, so all info it gives is 3 yrs out of date.
Fine for history or anything not time/date sensitive but forget any info on best funds for today
In it's current form perhaps not, but i'm very excited for what it might become.
Well, neither does a financial analyst isn’t? It can give you options …
...and next week's Euro millions numbers are..... D'oh!
I couldn't even get onto the website - its full.
Try in the mornings UK time, before the US wakes up.
Sri Lanka is an incredible country. Idve suggested some of the same.
Would be great for the legal industry.
Only if you have access to all the underlying data on which to base answers - which is not possible. Much of the source data is controlled by existing companies who are working on the same AI/ML solutions already.
I’ll clear out my desk
So this is basically Wikipedia 2.0 then?
Scary . End days beckon!
Be very very careful with chatGPT. It absolutely has zero concept of British tax regulations. Do not ask it for advice and help with your self assessment or anything else along those lines. It omits things and when you point it out it basically says "Oh yeah, so the answer is probably that then."
To quote Andrew Tate: 'the matrix has attacked me'.
If I could flood the internet with comments saying Where's Wally is the best book on wealth creation, that will be the book ChatGPT will recommend when asked. It is extremely sophisticated technology, but the ability to be gamed in that way suggests to me that there is a long way to go.
“Chatgpt is a data dump from 2021…”
Damn
Robert Kiyosaki has a reputation of being a scammer. That is not good advice from chat gp maybe?
I think is already more than capable to replace 111
Almost, if it’s got a 95% accuracy it’ll still be giving incorrect info to millions of people each year.
@@JamesShack interested to know human inaccuracy rate then...
If you can't come up with your own holiday ideas, then...
ChatGPT can't do what you think it can do.
So as AI gets more intelligent, humans will get more stupid.
Not really there’s 2 kinds of people
1) those who can extrapolate from incomplete data.
Yes indeed we have all the answers now on the internet but we are getting dumber.
Not necessarily
ChatGPT is superficially impressive - but it's not really an Artificial Intelligence - it doesn't really *understand* anything.
I agree although i think it will improve rapidly. How would you define, or test for, understanding?
@@JamesShack It will improve in simulating intelligence. It will never be intelligent. Simulating intelligence is all any algorithm in the much hyped and misnamed field of 'Artificial Intelligence' can ever do.
@@JamesShack haha. Good question. I guess that I'd look for it to have the ability to create a novel perspective - and for it to be opinionated. Ultimately, your question becomes a rabbit-hole. Like asking "what is consciousness?" Love the channel by the way.
Are we not just algorithms too?
@@JamesShack Science really does not understand intelligence. We know lots about the physics, biology and chemistry of single neurons but we know very little about how neurons act together to become conscious. It still seems almost like a miracle.
do think they want to serve 8 billion people without work? without money, and having freedom and nice life?
You need to emphasis bias programmed into ai as a potential negative.
its not 100% correct....................YET
Blocked here in Hong Kong lol
Hahahahahahaha… so you trust an AI solution run by a faceless corporate… “Tell me who I should invest my hard earned life savings in?…”. Yeah, I’ll trust that…!!!
Sounds like ai will make your job easier whilst making it more worthwhile for people to make use of your services thus you charging a higher fee, no?
Financial service fees have been coming down as technology and competition have proliferated. This technology will reduced the costs of providing advice enormously. So as long as the market is still competitive costs to consumers will reduce dramatically too.
Devils advocate. Financial advisors often too give incorrect information.
?
Dont lie, the bot wrote the entire script to this video didnt it?
Haha - parts…
This sounds like a cope.
Interesting question, I work healthcare performance and analytics. I feel like it would certainly help with the heavy lifting enabling more time in fine tuning. Funding that will be another matter 🫠, but that is another discussion not for this channel
I think businesses like OpenAI will develop into a Model as a Service business, leaving other companies to work out how to implement it for different use cases.
Asking ChatGPT for the best funds is like typing it into google and assuming the top 5 are the best. ChatGPT doesn't understand what makes a good fund, it just knows how people write about funds and makes some text up based on that. You be better building an AI stock model like I have, than using a language model. Predicting what you want to hear based on what other people have typed online isn't a good way to model finances lol
ChatGPT is just the text equivalent of a pretty new GUI. There’s no intelligence to the responses it provides - it’s just presenting the information in a more human-readable, natural language format. There are definitely many developments in Machine Learning / AI that are focused on predicting good investments - but these are nothing to do with ChatGPT.
ChatGPT could be the user interface that explains these investments to you in a way that a human would have done - but it’s still not the bit of code that’s making those financial decisions - it’s working out how to explain them to you in a way that feels like it’s human.