AI Financial Advisors: Benefits, Risks, and the Future of Advice

Name: The Pros and Cons of AI Financial Advisors
Uploaded: 2026-05-29T16:00:20+00:00
Duration: 19 min 26 s
Channel: The Plain Bagel
Description: Summary and key takeaways on The Pros and Cons of AI Financial Advisors: Summary & Key Takeaways, covering The Rise of AI in Finance OpenAI is piloting a

The Plain Bagel

May 29, 2026

•

19 min video

•

2 min read

YouTube video ID: mR4eOVo126Y

Source: YouTube video by The Plain Bagel — Watch original video

PDF

OpenAI is piloting a finance‑focused feature for US Pro users at $200 per month. The tool links to more than 12,000 financial institutions, pulls bank and investment data, and delivers expense breakdowns, portfolio reviews, and financial planning. Users can store “financial memories” to give context for future prompts, but OpenAI states the service is not a replacement for a professional advisor. Robinhood is also integrating AI trading agents, and existing products such as Portfolio Pilot already use ChatGPT to assist self‑directed investors.

Pros of AI Financial Advisors

AI tools expand access to high‑quality financial planning for smaller investors who are often excluded by traditional firms that require high minimum balances. They provide 24/7 on‑demand reviews, and many users report being more honest about their habits with an AI that offers no judgment. This honesty can reduce shame and lead to clearer financial pictures.

Cons and Risks

AI models are prone to “hallucinations” and can generate misinformation, a danger when advice involves money. The output is only as good as the input provided by the users, and prompt structure heavily influences results, creating inconsistency. Because the tools are not registered financial advisors, they carry no fiduciary duty, no “know your client” requirements, and no legal accountability for erroneous advice. Paid partnerships—such as Intuit’s integration—introduce potential conflicts of interest that may prioritize certain products over user needs.

The Future of Financial Advising

AI is unlikely to fully replace human advisors because it cannot manage the emotional factors that drive investor underperformance. Historically, many robo‑advisors have re‑added human components to satisfy client demand for peace of mind. The speaker recommends treating AI financial advice with the same skepticism as an anonymous Reddit post or a medical Google search. Regulatory clarity remains limited, and future litigation may shape how AI tools are governed in the financial sector.

Takeaways

OpenAI’s new finance‑focused ChatGPT for US Pro users costs $200 per month, links to over 12,000 institutions, and offers expense and portfolio analysis while warning it is not a substitute for a professional advisor.
AI financial tools increase accessibility by delivering 24/7 planning to smaller investors and encouraging more honest disclosures due to the lack of human judgment.
Hallucinations, dependence on user‑provided input, and sensitivity to prompt phrasing create factual risks, and the absence of fiduciary duty means no legal accountability for erroneous advice.
Emotional factors that drive investor underperformance remain beyond AI’s capabilities, prompting many robo‑advisors to re‑introduce human support for peace of mind.
Regulatory ambiguity and potential conflicts of interest, such as paid partnerships, suggest AI advice should be treated with the same skepticism as an anonymous Reddit post or a medical Google search.

Frequently Asked Questions

What are the main risks of using AI-driven financial advisors?

AI-driven financial advisors risk providing inaccurate information because models can hallucinate, they depend heavily on the completeness and quality of user input, and their advice lacks fiduciary responsibility, meaning there is no legal duty to act in the client’s best interest.

How long does OpenAI retain synced account data after a user disconnects?

OpenAI deletes synced account data within 30 days after a user disconnects the account, ensuring that the financial information is not retained indefinitely.

Who is The Plain Bagel on YouTube?

The Plain Bagel is a YouTube channel that publishes videos on a range of topics. Browse more summaries from this channel below.

Does this page include the full transcript of the video?

Yes, the full transcript for this video is available on this page. Click 'Show transcript' in the sidebar to read it.

Helpful resources related to this video

If you want to practice or explore the concepts discussed in the video, these commonly used tools may help.

Personal Finance Ledger For Budget Tracking Recommended

A physical ledger provides a secure, offline way to track expenses and financial data, reducing reliance on AI tools that may have privacy risks.

Amazon →

The Psychology Of Money Book

This book explores the emotional and behavioral aspects of investing, which the speaker identifies as the critical human element AI cannot replicate.

Amazon →

Hardcover Personal Finance Planning Journal

A dedicated journal helps users organize their financial goals and inputs manually, ensuring the quality of their financial planning is not dependent on AI prompt structure.

Amazon →

Links may be affiliate links. We only include resources that are genuinely relevant to the topic.

Summarize another video

Full Transcript YouTube

Hey everyone, it's Richard. You're
watching the plain bagel. Would you let
AI give you financial advice? That's a
question that's come up recently on the
news from OpenAI that CHBT will soon be
launching a new finance focused window
called finances specifically designed to
offer effectively AI financial advice. A
tool that will allow users to ask
ChachiBT for tips on cutting expenses,
guidance for their investment
portfolios, and even allow users to
throw together financial plans for
certain finance goals like buying a
home. giving people a lot of power to
carry out tasks that would traditionally
be carried out by a human financial
adviser. With this news being followed
up just this week with an announcement
from Robin Hood that the company would
actually soon allow users to trade
entirely through AI agents, signaling
for some a ramp up in the AI revolution
of the finance industry. Now, people
using AI for financial advice isn't
itself inherently anything new. Uh
people have long been making a habit of
asking platforms like catchupt for
things like stock picks and the like
with golic co-author of a recent
McKenzie report on AI and wealth
management highlighting that we're
already seeing over a third of consumers
across age groups using AI platforms to
help with investment guidance and open
AI themselves highlighting that more
than 200 million people come to ChachiBT
every month for budgeting questions
about their investments comparing
different paths planning for future
goals
and more. There are also already a lot
of finance focused AI offerings out
there that have built rappers around
platforms like chatbt. For example,
portfolio pilot uses chatbt to offer
quote complete financial advice for
self-directed investors. But this is the
first finance tool offered by Open AI
themselves. And what's different here is
that it will actually allow users to
connect their bank and investment
accounts directly to ChatBT so that the
platform can offer tailored financial
insights. with the announcement showing
a number of examples for how ChadBT can
utilize your personal data to inform its
financial responses. So, as a registered
financial adviser myself, I thought I'd
go over the topic of AI advisors and
whether people should use them, the pros
and cons they present, and generally if
they pose a real threat to advisers like
myself. Now, naturally, I'm going to
have a bit of bias here since uh I like
my job and would like to keep it, but I
actually think these tools might prove
quite useful if people understand them
within their context. So, I'll try to
present a number of different angles uh
to the topic, but we'll start off with
what we know so far about the chat GBT
finance offering. As highlighted by
OpenAI themselves, this feature is
currently being piloted to US Pro users,
meaning individuals who are paying the
$200 US a month subscription. So, it's
unclear if it's planned to be a paid for
feature or if it will be more widely
accessible with more basic tiers of the
subscription, but so far it's only
available at that $200 tier. And in
their announcement, Open Eye includes
all kinds of examples of how users could
get financial advice from the offering,
including asking for a review of your
investment portfolio, a breakdown of
your expenses by category and tips for
saving money, and other tailored
queries. With the platform able to
integrate visuals and charts into its
responses, and supporting more than
12,000 financial institutions,
demonstrating that this is likely to be
widely available regardless of where you
bank or or invest. The platform will
also store so-called financial memories
or details you share in prior
conversations to provide further context
for the AI model. So for example, if you
highlight that you have a specific goal,
the platform will save that information
to help with future prompts. In terms of
broader privacy, Chad highlights that
while it can access balances,
transactions, investments, and
liabilities, it does not have access to
account numbers or have any authority to
make changes to your accounts. with
Chhatcht giving users the option to
disconnect their accounts at which point
their synced account data will be
deleted in 30 days. However,
conversations with Chhatubt and
so-called financial memories will have
to be deleted manually by the user. Now,
OpenAI specifically highlights that this
is not a replacement for a professional
adviser, which I guess is their version
of #notfinancial advice. Uh, but you can
see it's trying to offer a pretty
comprehensive adviceike experience. I
know what you're thinking, Richard. Why
the hell would I trust my financial data
with an AI? Even beyond the more
apocalyptic concerns about AI using our
own data against us to take over the
world and enslave the human race,
finance is a pretty sensitive topic. So,
any advice on the subject is pretty
consequential,
and AI can't even spell at a fifth grade
level sometimes.
>> How many E's are in the word 17?
>> There are no E's in the word 17, even
though it sounds like it might. The
letters are s eve v e n t n. So it has
three e, not zero. I misspoke initially.
>> Well,
yeah, great point. As impressive as AI
models are, they are also so complex
because of their iterative self-coding
process that they're also prone to
hallucinations whereby they just kind of
start making stuff up, uh, spitting out
inaccuracies or misinformation as gospel
truths or just do more quirky things
like calling themselves Mecca Hitler or
bringing up goblins in their responses a
surprisingly frequent amount. However,
OpenAI tried to address that in its
announcement, highlighting that the
company had worked with a team of
finance professionals to train the model
and has benchmarked its ability to
respond to challenging finance tasks
with the latest model seeing a score of
82.5
out of 100, which isn't bad. That's a a
minus. Still, it leaves a pretty big
question mark about how bad that 17.5%
is. Is that just an error message or
rounding error? Or we're talking about
Chacht telling people to yolo their life
savings into Barkcoin. you know, the the
degree matters here. So, it's yet to be
seen how good or reliable this financial
advice proves to be, which might be why
we're only seeing this being piloted at
this point. But, as mentioned, AI
advisors are here. We're already seeing
a lot of offerings in the space, and
this just signals one more step towards
it becoming a more common offering.
So, what are the pros and cons? Well,
one of the biggest positives is that
these platforms make financial advice
much more accessible. Uh the truth is
the finance industry doesn't have a good
economic model for providing highquality
financial planning for smaller
investors. A standalone financial plan
can cost thousands of dollars. And
investment companies who offer highly
qualified advisors as part of their
investment services also tend to limit
their offerings to households with a
minimum account size, leaving most
people to deal with higher fee,
potentially lower qualified adviserss
from bank branches and the like. With
the eye advisor, however, people with
any asset base can get a pretty in-depth
review of their financial situation
247 on demand, which is a pretty big
positive. Now, the feature currently
only available under that $200 a month
tier. The ChadBt Advisor is still
relatively expensive on a $100,000
account. That would represent a 2.4%
annual fee if paid throughout the year.
But technically, people could just pay
for one month and get a, you know, get
all their prompts in. and people might
be getting value out of chat for other
reasons. So, you know, could still be
considered a fairly cost-effective
finance tool. Now, in addition to
accessibility, there's also the speed of
the response. The fact you can get a
financial plan within typically seconds.
Uh the fact that you can submit multiple
inquiries or prompts and uh request
different iterations of the financial
plan. So, there's a lot of flexibility
there. And there's also a chance that
people will actually be more honest with
an AI advisor than with a human one.
Finance, as you can imagine, is a pretty
vulnerable and sensitive subject for
some. So, people might choose to
withhold details from a human adviser
because of the shame, guilt, or just
embarrassment they feel discussing their
financial habits. Uh, you don't really
have that problem when dealing with
platforms like this. It's a lot like how
people are pretty willing to Google
their most intimate medical symptoms
before seeing a doctor because there's
not that fear that there's someone on
the other side of that screen judging
you for where you find yourself. So, as
you can see, there are some very valid
pros to the offering. But, as we touched
on, there are also some pretty important
cons with the big one, as mentioned
being that Tachibbt is sometimes just
wrong. As mentioned, finance is a
sensitive subject. The advice people get
can have a meaningful impact on the
quality of their life. So, while the
benchmarking data suggests that the
platform is generally good at giving
financial advice, it also seems to have
a bit of a gap. And those poor quality
outputs could prove harmful to people
depending on how severe those
shortcomings are. And while
hallucination rates are fairly low for
straightforward general knowledge
prompts with one metric rating the
hallucination rate at around 1% for GBT
5, which isn't even the current 5.5
model. It is fairly well known that the
more specific and specialized a prompt
gets, the more likely it is for
hallucinations to occur, which could
prove problematic when people are
looking for in-depth professional
financial plans. Another issue with AI
advisor is that the output is only as
good as the input provided by the users.
Yes, with the CHBT finance offering,
users will be able to link their
financial accounts uh which should help
better inform these financial plans. But
in terms of the prompts that users
submit, the details that CHAGBT has to
build these financial plans are limited
to whatever you provide it. Now maybe
Chad GBT will lay out the details needed
for these types of more extensive plans.
Uh if you prompt CHBT for example for a
comprehensive plan, it will list out
some information it requires. But
there's a good chance that these lists
miss some important details like whether
you're married with kids, your general
health, things that might come up more
naturally with conversations with an
adviser, but that users might forget to
include in their prompt when they aren't
given that more structured input
requirement. Another consideration here
is that prompt structure matters. You
could get different outputs from an AI
adviser depending on how you frame your
request. If for example I asked CHBT for
the top five stocks to buy today to hold
for 10 years versus the top five stocks
to buy for today's date, the list
actually changes slightly with one
including Alphabet and the other
including Amazon. Now that's not a
meaningful change, right? But neither is
the difference between the two prompts.
So hopefully you get my point. Someone
who isn't familiar with how to ideally
structure prompts could end up with a
subpar output for a financial plan.
There also privacy and security
considerations given how much sensitive
information people would be sharing with
an AI adviser and the vulnerability that
presents to either be hacked or even
abused by OpenAI themselves. And then
there's the fairly important point that
CHPT
isn't a registered adviser. Now I know
for some that might just sound like
gatekeeper talk, but there's a important
reason why that matters. As you're
probably aware, in the US and Canada and
most other developed countries for that
matter, you have to be registered as an
adviser to give financial advice. There
are some exemptions for things like
general advice where someone can go
online and generally talk about the
stocks they like and what they think
people should do with their money, so
long as it isn't tailored to an
individual's personal situation. But the
moment you start giving paid one-on-one
uh discussions or consultations
involving someone's money and how they
should manage it, you can find yourself
in hot water. But all this is exactly
what Chat GBT is doing. It's receiving a
fee to give financial advice. And while
the platform discloses that it's not a
substitute for professional financial
advice, that's not a magic barrier from
all legal liability any more than a
influencer putting #notfinancial advice
or prediction market putting
#notgambling.
Actually, apparently that one works, but
you get my point. Now, in the US, there
apparently is a route to provide
registered financial advice through AI
platforms. Portfolio Pilot, for example,
is an AI tool offered by Global
Predictions Inc., which itself is
registered as a registered investment
advisor with the SEC. But OpenAI has not
gone down that route. And even for
portfolio pilot, the registration still
required the CEO and chief compliance
officer to pass the required
examinations. So the platform argues
that the advice is entirely AI based.
The SEC is still interested in that
human knowledge requirement for those
involved. Now the reason why this
matters is that without registration,
OpenAI isn't held to the same rules and
requirements that registered financial
adviserss are. Meaning that there's no
fiduciary duty or duty of care, no
requirement to ensure that financial
plans are properly informed when they're
presented to individuals. Since there's
none of the know your product or know
your client requirements, advisors have
to demonstrate that they've done the
research to understand the funds or
securities they are putting clients into
and even things beyond that like in
terms of how to handle conflicts of
interest. For example, Chachbt announced
a partnership with in it the tax
planning software to integrate some of
its tools into the Chachib finance
offering effectively allowing
individuals to access certain tangible
tax tools and even professionals through
the Chachib platform. Now I have nothing
against in it. That actually sounds like
a good idea, but it is effectively a
paid partnership or promotion with in it
actually paying Chad GBT for this right
to integrate its tools. And this sort of
advertising could prove a very lucrative
revenue stream for OpenAI since they
have so much user data. People are
basically telling the platform what they
want and they can then connect those
people with advertisers who are willing
to pay for access to that market. But
that also comes with conflicts of
interest. If we start seeing those
finance tools integrated into this
offering, if for example, the platform
were to partner with an investment
company, maybe it would start
redirecting users to the investment
offerings, even if that's not inherently
what the user is looking for or if it's
not appropriate for their situation.
Now, to be clear, conflicts of interest
exist with professional financial advice
as well. Only about 3% of adviserss in
Canada are registered as fiduciaries,
meaning that they have to put the
client's interest above their own.
Meaning that most advisers are either
incentivized or limited in terms of the
solutions that they can offer their
clients. But how those conflicts of
interest are disclosed and managed are
pretty strictly regulated with there
being a lot of repercussions for
advisers who don't follow the rules or
again put their clients in inappropriate
products based on how they're
compensated. uh when it comes to ChadBt,
there doesn't seem to be those same user
protections. So, we'll have to wait and
see how regulators deal with this legal
ambiguity. I suspect that we'll see
someone eventually sue OpenAI for giving
inappropriate financial advice because
whether you're registered or not, uh
giving inappropriate financial advice
can still get you in trouble. But at
this moment, there's no explicit
protection for users. So, it really is
use at your own risk. Now, with all that
being said, I still think AI advisors
are a handy tool to have access to if
people recognize their limitations. For
DIY investors, I think it's awesome to
have another tool available. Although,
frankly, uh, for $200 a month, I think
there might be other tools that are
built specifically for financial
planning that would be better to pursue
if you know what you're looking for. Uh,
in the industry, for example, a lot of
advisers use software like Conquest,
which helps to build out those in-depth
financial plans. That one's not
specifically available to individuals to
use, but there are uh competitors that
you see for retail investors online. And
I think a more tailored solution like
that would be better in most cases if
you again know what you're looking for.
Yes, it seems chatbt is generally going
to offer good information most of the
time, but it's kind of like the jack of
all trades thing, master of none. Uh
usually better to go with something
that's been programmed with that
expertise baked into it. And given the
hallucination risk with LLMs, you're
going to want to scrutinize any output
from an AI advisor to verify the math
and reasoning before actually
implementing anything. Still, if you
treat ChatBT with the same skepticism as
an anonymous Reddit post or a Google
search regarding a medical concern, then
great. I think the more tools we have
for underserved investors, the better.
But you really need to understand the
limitations. For example, I really
wouldn't recommend using AI advisors for
things like stock picks, uh, despite, I
guess, having done so earlier. This was
a really popular thing to do when AI
first came out. People rushing uh, to
use the platforms to ask for stock picks
and, uh, research on different companies
and investment opportunities. And in
fact, we have platforms like GPT
investor that claim to have outperformed
the S&P 500 by relying on AI stock
picks. But when you scrutinize the
results, you see how unsophisticated it
really is. A lot of AI platforms, for
example, are yoloing their money into
individual stocks, which is why the
returns are higher is because you have a
few outliers pulling the average up. And
you have to recognize that LLMs like
CHACHBT are language prediction models.
They're not judgment models. These
systems aren't going through and
researching research reports, looking at
obscure data to try and come up with
some elaborate investment thesis around
these companies. It's effectively
putting together a string of words that
sounds like a good response to what you
asked it based on what it was trained
on. There's also the issue with stock
picks specifically that if AI is sending
the same five stocks to everyone who
asks it where they should invest that uh
you'd essentially be buying into an
inflated asset and the risk of that
being abused again whether by open AI by
hackers uh the issue of you know that
being taken advantage of. Now will AI
advisor replace typical financial
adviserss? Uh well, I do think that as
we get more DIY tools like AI that it's
allowing individuals to replace certain
advisor functions, but in its current
form, I don't really think it's a
meaningful substitute for advisers as a
whole. We saw a lot of the same
conversations when robo advisors were
first released uh with people thinking
that it spelled the death of financial
advisors because people could now on a
general DIY basis get guided asset
allocation for their investment
portfolios. And certainly the DIY
approach has been a growing trend given
the everexpanding toolkit available to
individuals. But financial advisors have
maintained their position in the finance
ecosystem largely because those specific
functions aren't necessarily why people
are reaching out to advisers in the
first place with one poll by Vanguard
finding that peace of mind remains the
top reason people seek out advisers. In
fact, some robo advisers actually moved
back to adding human advisers to their
offerings to better meet client demand.
AI doesn't offer a whole lot when it
comes to the emotion management side of
the business, which I think gets
discounted a lot, but is important. The
Dalbar studies, which have long argued
that most individual investors
underperform the general markets, often
cite emotional factors as being the
biggest drivers of this underperformance
rather than security selection or asset
allocation. So, while I do think that
we're going to continue to see that DIY
trend grow and advisers are going to
have to stay competitive with these uh
heightened offerings, especially in the
case where people might feel underserved
by their adviser, I don't think at this
stage it really substitutes the more
holistic advice that you can find in the
industry. Especially when you consider
how long it's going to take for users to
trust these platforms enough to rely on
their advice without feeling the need to
scrutinize or verify everything.
>> Uh, okay. I see what you mean now. Yes,
there's only one E in 17.
>> But then again, maybe that's just cope
since I'd be out of the job if that
happened. So, you can let me know in the
comments uh what you think about AI
advisers, if they're really a
replacement for traditional advisers. Uh
thanks for watching. I hope you found
this video helpful. If you did, please
do make sure to like, subscribe, all
that good stuff. It does help the
channel tremendously. If you like to
hear me talk more about this exact
topic, I recently filmed an episode over
on the Dead on the Money podcast with my
co-host Derek Deadman, another
registered portfolio manager on this
exact thing. I did steal some of the
data points, so there's a bit of
overlap, but you can check it out if
you'd like more content.