Topic: Chat GPT do a TA about bitcoin - page 2. (Read 573 times)

It's interesting to see that how AI Chat GPT help you but in my mind you and everyone should avoid it because here huge kind of risk involved, as from different people review there are alot of mistake done dy Chat GPT in different aspects and still ongoing so doing Technical Analysis from AI will cost you too expensive Because it's not like a human who will be responsible for any kind of loss through his Technical Analysis .
As for better Technical Analysis you have to done it by yourself   because let suppose you lose money by your own Analysis will not hurt you much as because on the others hand it will teach you a lesson and help you to gain experience.

I've copied the daily OHLC from coinmarketcap and pasted into the discussion. GPT chat has no problems remembering what user said earlier in the conversation ... or at least that's what openAI claims, in fact it wasn't that easy, because the bot got lost when combining data from several messages, it didn't inform that some of the messages were lost by character limits, etc.

Yes of course, it will mention that it does not have data for future events also, so any information it's providing for bitcoin future price is purely speculation.

Please, how were you able to provide it with current data? Please share with me. Thanks!

It's a fair point. There are also clear differences from crypto influencers posting their TA online, as a lot of their TA is bias based on what will give them clicks, likes and more clout. At least ChatGPT doesn't care about clicks and likes, or it's reputation. When it comes to many subjects ChatGPT will give you dumb answers, but when it comes to analysing data based on historic techniques, it's probably not far off accurate.

After all technical analysis isn't looking into a crystal like many people assume, it's merely applying probabilities to different scenarios, generally based on technical indicators that have been around for years or even decades. I'm sure in the near future there will be more AI-based trading bots, given that there already are success trading bots based on specific indicator-based strategies.

On another note, can you ask it more specific questions, such as what is the % chance of the 200 Week MA holding as support, or otherwise price re-testing $20K, or breaking $30K, etc? I mean rather than simply short, medium and long-term price predictions, more price-specific questions. That'd be interesting to hear what ChatGPT considers the % chances of reaching certain prices are...

The fact that when you asked the AI bot a question and returned an answer of only being able to process data up until 2021 makes it a waste of time for me!
But of course for other questions they do give satisfactory answers, and this being an era which depends heavily on such technologies I wouldn't want to be fighting this tech but embrace it with due diligence!!!

For those that might want to be reliant on this tech , try ask AI to solve mathematical questions and see it fail or bring out wrong answers...which is why I say proceed with caution!

Btw those AI predictions arent convincing for me because its being spoon fade with the kind of data it needs to generate an answer which makes it less authentic for me

How does it differs from us? how does it differs from every single crypto celebritiy posting their analysis on twitter? why do you rank the human brain so high compared to AI? the truth is that both the AI ​​and the human brain only do one thing - desperately searches for patterns in the soup of information it receives. the answer to who will create a better analysis, human or AI, depends only on how much data in what time and how well it is able to process and how good a pattern to find. In the first 2 indicators, AI definitely beats us. It's only a matter of time until he starts repetitively winning in the third. In my opinion, it is worth learning how to use AI now because traditional information-finding will soon become ineffective in such a competitive profession as an investor/trader.

Saying "why should I use an AI when I can do it myself" is like saying "why should I use a computer when I can count everything myself on a piece of paper".

To be honest chat GPT is just a bot, which uses information available all across the web and then gives a detailed summarised answer based on this. Moreover many times it has provided inaccurate answers to some prompts also. So yes regarding investment plans, then I won’t suggest any bots for this. The TA that ChatGPT has done can also be done by you and that too more accurately if you have visited few good websites. Nevertheless ChatGPT is just another bot which makes small tasks easier.

ChatGPT cannot be accurate in predicting bitcoin market price because bitcoin itself is volatile, earlier today, bitcoin worth about $27,200 and later dropped to $26,800 this is a clear indication that we can have more high or dip anytime soon, and the trend is more likely to focus on going high, but if we can realize the breaking limit of $30,000 this new month then we may be getting closer to more rise as we continue to approach halving, let's see what this month has to offer with the market price.

ChatGPT definitely knows how to tell us what we want to hear. We are all bullish in the long term and bearish in the short term, when it comes to predicting the Bitcoin price. I don't need AI tools like ChatGPT to tell me what I already know.
It's funny how some people think that such AI tools can be viewed as some kind of Oracle or a prophet.
I use ChatGPT only for fun. There's no reason to take seriously the inaccurate answers of some imperfect piece of artificial intellect.

Yeah interesting but large pinch of salt isn't going to be enough. I would take it with one wheelbarrow full of salt. There's several screen grabs where ChatGPT has outright lied. I can't use a term "made a mistake", because it seem to invent some things, never heard persons that don't seem to even exist and things that don't make any sense.

But i would like to know what chatgpt 4 had to say about this as it does have access to net. It could even code a prediction program. Maybe Chatgpt could give an algorithm how it predicted this one as well.

That's great if they said that but I think there are still people who can't accept that fact. They are still expecting more from these AI's especially now that there are new AI's like Chatgpt and it was also being improved from time to time. Right now, Chatgpt is on it's version 4 but its performance are still kinda bad. Just take a look at the response get by @bettercrypto. What surprises me is he still call Chatgpt good.

You are right that they are delayed. Using them is not only a waste of time but its also a waste of money. This is why it's only better to rely on our own analysis. It may not be the best but at least it is more updated than what the AI's has to offer.

your description of Ai is good. Transferring the knowledge of how it works with mario to the language model, it looks like it is filled with a lot of filtered human conversations and texts written by people (the source of the GPT chat data is secret). As a curiosity, I will say that he also has many books in his "knowledge" because I had interesting philosophical conversations with him based on the works of, for example, Harrari. All this knowledge, however, is useless, because the algorithm does not know what is good and what is bad This is where tedious long-term and expensive AI training comes in. The AI ​​gets a question, creates an answer and gets feedback from the user whether the answer was right or wrong. At first, wrong answers prevail because the AI ​​is just "learning" but with time the quality of the answer increases.
that's why openAI has made its tools available for free. They incurred huge costs in data filtering and initial training, pay for computing power and would share it for free? They do it because every user helps with the training. OpenAI even asks for as much feedback as possible while creating an account. input data filtering is extremely important. Hence connecting ai to the Internet is not as easy as it might seem. Immediately our algorithm is exposed to data spam attacks from hostile AI (garbage in garbage out)

Combination of both. The main way how an "AI" works right now, is that it is fed with information, and it learns what equals what. I always give the same mario example, there is a video on youtube if you search for it, it's a machine learning that plays mario and finishes it eventually. You feed it needed information, like what button is jumping, which creatures makes you die, and what is your purpose etc, so the game rules basically, and let it be, then it fails after and after again until it figures it out the rest itself.

So yes, it does think for itself in a manner, but within the box you put it in, so you give it information, and then you ask it a question and it uses the information to respond, I don't know if I made it clear or is it still confusing.

It has recent price data because i loaded it with OHLC values of last 6 months of trading by myself.

Why should anyone care about ChatGPT preditions if there are more than enough analysis created by human? Well i think that correct answer to this is an approach devoid of emotion, conviction and interest in persuading you to buy/sell. The analysis is being created on demand and you can at any time ask an additional factor to be included in the analysis (e.g. RSI or the influence of moon phases, whatever you want), ask for something that interests you in the analysis or ask for clarification of something you did not understand in the analysis. You will receive an updated analysis with explanations in 5 seconds.

I don’t know, man. I’m so confused reading about AI because I don’t get one thing: is it actually able to logically assemble information or does it just find combinations of words and phrases that generally make sens gramatically and assemble it all as a string of text?

Asking because it seems so weird reading AI TA. Like are we reading some text that made an actual logical sense to AI or what is this? Am I the only one who doesn’t understand? I kinda look at it at some kind of more advanced text spinner, but nothing else..

This is somehow interesting though because it analysis is almost related to what is currently happening in the market right now, I wonder how it fetched these recent datas that it supplied you since it's not being connected to the internet to source out this datas, however there's no special work that has done here by the ChatGPT differently or more than the way we humans has been analysing the market that I should rely on the data from ChatGPT when I can read and understand what is going on there in the chart plus the news .

Really??? As simple as your English is? Common!!! Below is your quote that I replied initially to:

And from the quote, only how we see the response to your ChatGPT query and our thoughts about AI in relation to TA are the only two questions that could be drawn from it. Now, going back to my first post, you would realize that I said, "If the AI has the information and capability to read the future price of the market, it won't still be productive, do you know why? It's because it will always rely only on the data at its disposal, but the market is too dynamic for that." That is the response to the first question in case you didn't read it insightfully.
 
While my general thought on AI, which is the second question on this subject matter is the rest of the body. This encompasses what present AI could do and how cautious developers have to be to avoid creating monsters in the market, which is where the violation you replied to first came in. It's part of my thought, it's an open expression since you asked for our thought.

And in case you didn't know, if AI works perfectly, it will be a matter of time before everybody knows how, and be getting rich. What do you think would happen to the market?

My dear, the bottom line is that AIs will try based on the information they have just as I iterated earlier, but still, they will fail or have very low accuracy. While developers will continue to be careful so that they don't still violate rules in coding them.

Are you clear again?

So you obviously don't understand what the thread is about. The thread is about whether and how current language models are able to model bitcoin pricing based on data provided to them. This topic is definitely not about cybersecurity and the philosophy of artificial intelligence development.

In addition, trading algorithms are mostly based on simple mathematical operations (for example if BTC/USDT * ETH/BTC > ETH/USDT than buy ETH with USDT, Sell to ETH to BTC and SELL BTC back to USDT with tiny profit) and have nothing to do with artificial intelligence or machine learning which are a definite minority.

"Are you clear?"

You might want to read my above quote once again, maybe if read with more understanding might tell you something, and that you are out of the line. I've noticed one thing about you guys, you either take the word out of context or you don't seem to understand what you are replying to.

For clarity's sake, I know that AIs are being used to trade, are they not trading robots or bots? But when it comes to such as ChatGPT, then the advancement comes, and so is the care so that it will not intrude into the financial market trading affair through hacking and other dangerous vices. This is why the developers of such AIs tread carefully, especially with the sector that deals with money so that they don't make a monster along the line.

Are you clear?

As I already said. The chat gpt data ends in 2021, but i loaded it with fresh OHLC data from the last 6 months of bitcoin trading. So chatGPT had trading data up to 2021 and from Jan 2023 up to today. So he knew where the previous ATH was, where the last bitcoin bottom was and price action of last 6 months.