Topic: I asked Meta AI two questions about bitcoin (Read 355 times)

The developers cannot possibly programme the software with all possible answers and keep updating that. There are a gazzilion possible questions someone can as AI and the answers can only be gotten from crawling the internet.
Not all questions are facts, some are subjective so there can be different possible corrections and the AI will choose the majority opinion all of the time.

Search up what AD means and all the results say it is Anno Domini, that's where the software gets its answers. It's only commonly used as "After death" in conversations, but almost all research articles term it correctly.

Which doesn't happen when you ask questions that will force the AI to search for subjective and biased information. After the recent massive hype, many people is disappointed with the results to questions like these.

AI is great for other tasks like lab research, creation of drafts,... but when you need more complex answers to questions where critical thinking comes into play, the current results are quite useless, as the OP shows.

I am with you on this, I have known about the fact that AI often messes up with their answer especially when it relies on the recent information that is circulating the internet.  Since there are lots of fake news and misinformation lurking on the internet, any inquiry on AI application that relies directly on the internet will more likely give us mixed information including fake news and wrong information.  And if we dig deeply into the question, this kind of AI will eventually admit that its knowledge is limited and the information it gives maybe wrong.

I think it is much better to use an AI that is fed with verified information and stored in a database, although it might be lagging in updates, at least any question asked has a high chance of being answered correctly.


True that since most AI today's rely on the internet with lots of fake information, using this kind of AI may inevitably spread wrong information to those who are seeking knowledge but are lazy to verify the information they gathered.


It is a no-brainer that people rely on AI for convenience and speed.  It can easily retrieve information from a database and make a piece of complete information regarding the subject matter.  As long as the saved inputs on the database are correct and verified, it will speed up any research and activities several times.

For those who still rely on AI, from this topic the answer can be found that relying on AI can be "dangerous" in terms of finding information because the truth of the information provided can be misleading. Regardless of the form of questions asked by the user.

I am confused why people who are active here still rely on this technology. I mean, AI technology can indeed be helped and I will not ignore it, but not for all lines.

You shouldn't be disappointed because these softwares are being developed by human who are bound to make mistakes. Much should not be expected from it and moreover, if you check their term of service there is disclaimer notice that the software may make mistakes and one should not rely on every information they provided.

I think they are just save for risk free and less important information that is not related to finance. They are meant to provide simple and quick guide in case you have no means to access the right source, they will give clue to what you are looking for. However, the technology is just developing at their early stage and would probably do better as they advance.

You're missing something here making things easier is just similar to improving humanity.

Ai ought not to stand with majority but what's right the dev ought to give them the right data or at least create them in a way they could remember after they're corrected.

Let me give an example, in terms of Calendar many takes AD as After death whereas it means Anno Domini I bet when you ask Ai they'll give Anno Domini as the correct one simply because the dev gave them the right data there for it. See another reason here, why blameis on the dev

Perhaps the goal of AI development is precisely that people become as less creative as possible, that they think as little as possible and that in the end they do exactly what the AI ​​advises them. I personally see it as one of the ways to dehumanize human society in addition to everything that is already present in so-called modern societies.

I can't remember any case in the science fiction where AI and ultimately humanoid robots powered by AI doing anything good for humans - on the contrary, it always ends up very bad. Humans will obviously not rest until they invent something that will be superior to humans, but if you give someone consciousness, then it is only a matter of time when humans will be perceived by those same creatures as a lower species.

The use of AI is something that has already been programmed by some set of people whose pattern of the informations they give were tuned to the conformity of their own taste and standard, this also make it a centralized source of an information in which their opinion, suggestion or given answers may not be accurate, the use of AI is not 100% precise of the exact information or response needed, for those using them, they have their own reasons, but in case of bitcoin, it's more better not to involve the use of AI in any research of information we needed.

Unfortunately, the crypto content on the Internet is bad, and most of the sites that add learning algorithms are bad. If the answers are taken from books, technical forums, and a little content from here, you will get better answers, but since the learning sources are bad, the answers will inevitably be bad.

The thing about AI is that its as good as its source data, and probably this information is presented to us depending on how it ranks on search engines... as AI is just a simplified version of any search engine out there such as Google.
Which is why I tell my friends to seek second opinions from other AIs or go back to googling so that you pick your option out there.

Btw, something I appreciate from the AI used here is that it gives you the link it's pulled it's information from unlike others that will just list stuff in bullet points and call it a day... otherwise not a bad start from Meta probably will get better with time.

Or probably this technology is fairly young and still way beyond what we expect of them. But the thing is that there are social medias who really portray AI as the know it all and that it will take over our traditional jobs.

But this thread just shows that it's not going to happen maybe not in our lifetime as it still needs human to feed it with information and most of the time it's going to messed it up, hence it's not the answers that we wanted to see or hear.

I won't be surprised to see those information that is generated from AI and of course these information are gotten from the website that doesn't give accurate information concerning what they want the AI to passed to. In my opinion AI may mislead lot of people who doesn't have the right knowledge towards bitcoin especially the newbies who eagerly to learn and delv into bitcoin investment.

They were invented to make things easier not to improve humanity. Artificial intelligence functions by learning from us and refining answers, so when you ask AI to give you a good R&B beat, the AI does not start singing from heart but checks what humans consider sch best to be, collates lots of answers and gives you the best possible one it can compile, if the general idea of what such a best should be, AI cannot correct that.

Those concerned about privacy, self custody anv verifiable softwares make up the minority still.

Google provides fake information? or Google's AI provide fake information. Information is provided by the media providers like content managers or writers or web masters etc. Google's algorithm just show you what you want to see according to the nature of your keywords. It never showed fake information to me. I would love to hear if it does to you if you don't mind.

Overall AI is not fully accurate especially of WhatsApp AI. I have tried it and my experience with it was not so good. You should try Gemini or GPT or Claudia they are good and accurate. You could also test Microsoft's Copilot AI as it also provides good information but sometimes all these AI provide wrong answers to the questions you ask. Actually either they are outdated or you did not asked the write question. Like the future you wanted in some wallet if you would have more precise with your question it will understand you better and provide you a better answer.

There is a field known as command prompt where you learn how to give command to an AI bot and there are whole lot of courses on this single field people are giving input to AI in tens of line to get accurate answer. So yeah I think AI did not reach to the point where they will give you exact answer while you are giving just few words input. They are not sync with your mind yet.

Sorry I got it mixed up in my head earlier wanted to say they're fond of listing closed source non custodian wallets and the other lower tier like Mycelium though which is  open source.

The fact that majority are wrong doesn't means Ai will also follow suites after all they were created to improve humanity but unfortunately they have failed in some key areas, a clear example is  this and the dev also still has a take in this blame.

Not all Ai are built to learn from the context of the discussion with users or may hold on to the correction based, example is the Google Gemini Ai base on this reply here and I think the WhatsApp Ai is no exception on this.

AI can be very inefficient as they most times provide wrong answers to a particular question we ask. Well, we can't really blame an AI for been inefficient, and here is my reason: AI carry out tasks such as answering of questions, by looking through every data available on its database. The data that are available are mostly search results we can find ourselves on the internet, which means the reply that was giving to you by the AI was literally writing by a human, but was available for the AI to use. AI just try to put every piece of this information together inorder to act efficiently.

It would be a big problem for those who depend only on AI results, and have no idea of platforms like BTT. Accepting most of this AI content finds a way to lead to misconception, and sometimes create unnecessary arguments or debate. Let's try as much as possible to be guided, or better still ask relevant questions on the forum.

Just as Lucius as stated above, I think humans are underestimating their own capabilities. It's so sad that some people have put about 99.99% trust on this technology, allowing AI to do our reasoning and other sh*t.

People overestimate the capabilities of today's AI and think that they will get better answers from it compared to asking the same question to a human. This is not only a waste of time, but also an underestimation of human beings, whose brain is the most intelligent computer in the world, but the less we use it, the more stupid we will become, but we have AI that will think for us, right?

The way I see it is that it works like a search engine but scrapes compatible datas, patch it together  to give a reply, if you try to correct it and give it a more appropriate answer, it's likely that it heed to your answer or still disagree , tried that with Gemini sometimes  when I correct him for some information  given he refuse it and insist  on his answer especially  when it's mathematically  related . The funny thing is that at the moment it heeds to your answer if you ask it again it will give your answer sometime but if you tryit again maybe after a day it gives you the same answer different from yours again...

So don't trust AI too much. This is really dangerous, I understand they work from big data, but they have the data used not having credibility. But maybe this is also influenced by the many writings about names or sentences that appear from AI applications. Their filters can't seem to figure it out very well. They work from the majority of the words or sentences they collect. This is the weakness of AI, I don't have high hopes for this app. 

For the most part they don't get fed data, they are programmed to get the data for any questions they are asked from the internet, just as you will do if want to get them yourself. The AI is not objectively wrong based on societies standards of what the answers to those questions are, do a Google search for "best Bitcoin wallet", those listed will be at the top of the search results, so someone who doesn't know where to look will get those answers and run with them.

AI gives you back the most popular results, so Coinbase and trustwallet are used more than electrum, blue wallet and Bitcoin core and majority of people consider them to be the best.
On play store, electrum has 1M downloads, trust wallet has 10M and a higher rating 4.4(from 1M reviews) to 4.0(from 3K reviews) for electrum.

This is not a problem with AI but a reflection of how society view and use Bitcoin.

All wallet listed are noncustodial but you are right about the AI given wrong information about bitcoin which is the reason I sent that merit. I will stop using it for bitcoin related things. All of them are disappointing like Google as usual.


The wallet that was listed that I posted above are all noncustodial wallet. The Coinbase it was referring to is the Coinbase noncustodial wallet.

But as I know that it will disappoint me again, I asked Meta AI your question and this is the answer it gave me and a link as usual:



You can see many wallets that experienced bitcoin users can not recommend listed among.

The question is not clear so the answer can be too general.

Best wallet, on what criteria, because different people will find their own best wallets. You can have your needs and see this wallet is best for your needs but the others can have different demands so their best wallets might be different than your choice.

How Do I Choose a Wallet?

This saying is true "No matter what choice you make, you will obviously be criticized by almost everyone who did not made that choice"

If the question more focuses on privacy or coin control, the answer from AI can be different. People needs to know what they ask like if a newbie, they don't know too much about many things. If they have demand, are unclear, they will ask more questions and AI can help them with more details.

Charles-Tim can try another question, and see what is AI answer for it.

It is not wrong on the number two questions though as you just asked it the "best wallet", it answers with different type of wallets, and its true that many uses coinbase and trustwallet and maybe considered as "best". Try to ask different question like "best non-custodial wallet" or "best custodial wallet" surely, it will give a different list of answer.

A lot of people still don't really get what large language models like Meta AI can actually do these days.  These LLMs are trained using large stacks of text and code, but they can still mess up or say stuff that's not quite right, 'specially with complicated topics.  Like you saw, Meta AI probably just linked you to where it got its info from showing it has limits.  It cant check out or break down info it finds on its own.  I'd say it's best for communication and summing stuff up, but AIs are nowhere near perfect when it comes to facts.

I tried this meta AI once just out of curiosity and it give me an answer exactly from a website and the link to the website exactly as it was in your screenshot.
It's obvious that the AI doesn't really have data of its own. It's just Google-like, as you said. If you ask a question it would give you a link to that that question. Problem with this is that the website is owned by people and people can be wrong or have a different perspective of things. Also, most websites don't get updated on things like this, so you'll be given outdated data in most cases.
Personally, I haven't seen the problem a simple Google research won't solve for me that chat AI will.

Actually getting information about cryptocurrencies especially bitcoin right from Ai will be among the most easiest ways to see in the flaws in them, I wonder what kind of data they feed them to make them like this I guessed this goes down to the dev.

It's usually common though to see Ai listing custodian and other low rated  wallet as best wallet but I'm not surprised after all Google search still do just that and I guessed they may be getting info from there.

This is my first question which is what is bitcoin segwit address:



It only gave me some answers and gave me the link to the site it got the information from. I have not seen segwit versus native segwit before. It is nested segwit versus native segwit which is correct. Nested segwit can be referred to as compatible. Nested segwit is not segwit. The native segwit and taproot are segwit. Meta AI gave me a link to information that is not accurate.

That one is even better. The second question is more full of wrong information.


This is my second question which is about best bitcoin wallet:



Most of the wallet listed are one of the wallets that should be avoided except Trezor and Mycelium. Mycelium is not even having replace-by-fee and coin control but it is manageable as it is open source.

All of them are not more than Google-like with fake information. I am disappointed to try this the first time to know that it can mislead people.