Stocks go up and stocks go down.
If you could somehow accurately predict when they are going up and when they are going down, you could be an extremely rich person. The problem is that making accurate predictions about stock prices is a long-time elusive goal. There is an old adage that mockingly says it is difficult to make apt predictions, particularly with regard to the future.
You could try to make predictions by using your gut instinct. Many people do. Another approach consists of trying to employ the most sophisticated mathematical models and using the zenith in calculus to figure out where stock prices are going to go. All manner of apps are available to aid you in your quest to forecast stocks.
Into the hustle and bustle of the stock advisory mix comes the advent of Artificial Intelligence (AI).
In today’s column, I will focus on how generative AI is being used to predict stock prices. Generative AI is the latest and hottest form of AI. There are various kinds of generative AI, such as some that are text-to-text based, while others are text-to-video or text-to-image in their capabilities. As I have predicted in a prior column, we are heading toward generative AI that is fully multi-modal and incorporates features for doing text-to-anything or insiders say is text-to-X, see my coverage at the link here.
In terms of text-to-text generative AI, you’ve likely used or almost certainly know something about ChatGPT by AI maker OpenAI which allows you to enter a text prompt and get a generated essay in response. For my elaboration on how this works see the link here. The usual approach to using ChatGPT or other similar generative AI is to engage in an interactive dialogue or conversation with the AI. Doing so is admittedly a bit amazing and at times startling at the seemingly fluent nature of those AI-fostered discussions that can occur.
The question that might be in your mind is whether or not ChatGPT can be used to predict stock prices.
Yes, it can, though I will point out that there are many pitfalls and considerations that you must keep in mind and I will seek to so enlighten you in this discussion herein.
MORE FROMFORBES ADVISOR
It might seem at first glance to be counterintuitive that ChatGPT could be used for stock price predictions. You see, ChatGPT is mainly a text-to-text or text-to-essay generator and therefore would not seem especially suited to dealing with the numeric realm of stock prices. There are many well-known issues associated with ChatGPT trying to do everyday numeric calculations accurately. You would be on shaky ground to try and natively have ChatGPT do stock price analyses for you.
You could seek to augment ChatGPT with other allied apps that do in fact provide detailed numerical calculation capabilities. Recent announcements associated with ChatGPT have noted the availability of other apps that can work hand-in-hand with ChatGPT and take on the chore of numerical calculations, see for example my coverage at the link here.
There is another avenue available that entails pretty much using ChatGPT all by itself as a viable means of doing stock price predictions. That being said, you don’t have to use ChatGPT on a standalone basis and there is no particular compelling reason to avoid augmenting ChatGPT, but, in any case, let’s for now focus on the solitary use of ChatGPT in the pursuit of hoped-for wealth building.
How would ChatGPT be useful for stock predictions?
The answer arises in the richness of the text.
Think of it this way.
We can reasonably assume that stock prices are to some extent shaped by what is being said about a stock. If a banner headline in the news is blaring that a stock is going to go up, the chances are that it will go up. This seems sensible in that people will read the headlines and opt to buy the stock and ergo push the price higher and higher. Likewise, if headlines trounce a stock, the odds are that people will flock to sell the stock and get out of it. The stock is undoubtedly going to drop in price.
In short, by examining the headlines about stocks, we could presumably predict which way a stock is likely to head. There isn’t an ironclad guarantee on this. Headlines with a positive message or sentiment about a stock could suggest a rise in the stock, but there is also the possibility that people will ignore the headline, or they might act contrary to the headline, etc.
Trying to rely on a singular headline alone is a bit dicey. Nonetheless, it can be a powerful tool. You might also opt to look at lots of headlines to gauge an overall trend. You might want to examine headlines over a period of time, rather than relying on a moment in time. A slew of clever ploys can be used that make use of headlines as a leading indicator about the direction of a stock.
The crux here is that a headline is composed of text. We can thusly use a text-oriented AI tool to examine headlines. And, most importantly for today’s discussion, we can in particular use generative AI to review headlines and attempt to make stock price predictions, including for example by using ChatGPT.
A common parlance is that via AI we can undertake sentiment analysis of text associated with whatever we are trying to predict or assess. The Natural Language Processing (NLP) facility of AI is able to parse words and attempt to glean what sentiment or expression is found within that text. You could of course ask humans to do the same thing. Indeed, there are specialists or stock tipsters that keep track of headlines, and they will tell you what they think those headlines will do to various stocks.
Generative AI such as ChatGPT is being used at times in conjunction with stock tipsters or lieu of them. It all depends.
Individuals that are doing stock trading are turning to generative AI to give them another angle on sentiment analysis associated with stocks. You might choose to do so and seek no other inputs. Or you might use ChatGPT and then compare the results to other akin indications. A variety of combinations and permutations exist.
I’ve so far emphasized the use of headlines for undertaking sentiment analysis. Headlines are handy and usually easily obtained and examined. There are other possibilities to be considered. For example, you might examine social media postings such as tweets. Those are again text-oriented and therefore readily examined via generative AI such as ChatGPT. I’d like to add that there are a lot of generative AI apps these days and you don’t need to confine yourself to only using ChatGPT. I am mainly mentioning ChatGPT because it is widely and wildly popular, said to be the 600-pound gorilla of modern-day generative AI.
There are stock-related headlines to be explored and tweets to be studied.
What else might we look at that is text-based?
News stories would be another useful source. The issue with news stories is that they involve a lengthier chunk of text than a headline or a tweet. A human-based analysis is accordingly more labor intensive when dissecting an entire news story. The beauty of using generative AI is that the longer text is not a big deal per se, other than the computational processing required to do the sentiment analysis (this can be an added cost, possibly notable if you are paying for online access to your generative AI app).
Various types of text-oriented content can be used when seeking to do sentiment analysis, such as these types:
- Sentiment analysis of headlines
- Sentiment analysis of social media such as tweets
- Sentiment analysis of news stories
- Sentiment analysis of firm-specific released info
- Sentiment analysis of governmental reporting
- Sentiment analysis of widespread polls
- Sentiment analysis of written expert opinions
- Etc.
Another aspect to consider is what level of sentiment analysis you want to perform. You won’t necessarily be able to find text that pertains to a specific stock of interest. Instead, you might need to look at a broader scope of text that is impactful to the stock you are focusing on.
Here is various level of text associated with doing sentiment analysis:
- Sentiment analysis of the global economy
- Sentiment analysis of a regional economy
- Sentiment analysis of a national economy
- Sentiment analysis of a local economy
- Sentiment analysis of a particular industry
- Sentiment analysis of an industry segment
- Sentiment analysis for a specific firm
- Etc.
You can likely discern that this is not a one-size-fits-all circ*mstance of using generative AI for generating stock advice. There is a plethora of options to consider.
I mention this to bring up a vital point.
Here it is:
- Be extraordinarily cautious and mindful of using generative AI such as ChatGPT for your stock-picking advice, and do not fall for the emerging and likely expanding scams that will make outrageous claims of vast riches by using their proffered ChatGPT prompts and instructions.
Yes, there is a cottage industry growing of those insisting that they can tell you just the right kinds of generative AI prompts and approaches to turn ChatGPT and other generative AI into your stock-picking guru. All you need to do is pay them some bucks, and they will reveal the inner secrets of how to get ChatGPT to guide you into stock-picking nirvana.
This is a sneaky twist to the classic stock picking get-rich-quick sells pitch. The new approach leverages the excitement and mystique of generative AI. Tossed into this is the ever-useful FOMO (fear of missing out). Everyone else is seemingly going to use ChatGPT and generative AI to select their stocks or advise them about what stocks to trade. You, meanwhile, are going to be left in the dust. Wake up and smell the roses, you are told, generative AI is where things are at to be on your way to immense wealth via stocks.
Please be on your guard.
I’ll try to arm you with sufficient knowledge to be on top of what makes sense and what is blarney when it comes to this topic. Benjamin Franklin famously said this: "An investment in knowledge pays the best interest." So be it.
The arena of generative AI is gradually going to have AI apps that are specially crafted for stock advice generation.
I divide the arena into these three realms:
- 1) Generic generative AI. General and widely used generative AI such as ChatGPT accessed to perform stock analyses and make stock price predictions
- 2) Stocks-Tuned generative AI. A vendor takes a generic generative AI and augments it with plugins or add-ons to hone toward generating stock analyses and stock price predictions
- 3) Stocks-Customized generative AI. A vendor makes a customized generative AI that is solely aimed to perform stock analyses and make stock price predictions
I’ll say more about plugins and add-ons for ChatGPT and generative AI in a moment.
Another consideration entails whether you want the generative AI to merely give you advice or whether you want the AI app to also perform stock trading for you. The usual approach consists of asking ChatGPT to generate an essay or response to your prompt, of which you then decide to proceed to do stock trading or opt to not do so based on what ChatGPT has indicated to you. You are in the loop.
The sketchier approach consists of wiring up ChatGPT or whichever generative AI is being used and you allow the AI app to undertake a stock trade on your behalf. In a sense, you are no longer in the loop. You are allowing an AI app to automatically proceed ahead. Be aware of the dangers associated with that kind of automatic setup. Sure, you might have also identified guardrails about what can be done, nonetheless it is presumably going to be acting without your explicit human intervention and moment-to-moment approval.
There are these two major ways to proceed:
- a) Advisory only. Generative AI indicates suggestions about stocks and stock trading, which the human user would potentially use to make actual stock trades
- b) Plugged Into Stock Trading. Generative AI that is hooked up to stock trading systems and can perform stock trades without the need for human user intervention
I will walk you through an example of using ChatGPT on an advisory basis. The same setup could be adjusted to include a plugin to actively proceed to do stock trading.
A few added caveats before we get further into this.
I’ve previously covered the various prohibited uses of ChatGPT, as stated by OpenAI in their licensing stipulations when you make use of ChatGPT, see my review at the link here. Included in the not allowed list is this: “Offering tailored financial advice without a qualified person reviewing the information. OpenAI’s models are not fine-tuned to provide financial advice. You should not rely on our models as a sole source of financial advice.”
There is a murkiness as to how this usage policy pertains to using ChatGPT for stock advice and/or for directly connecting to stock trading add-ons.
I have also discussed that ChatGPT and other generative AI are leaky regarding private data that you might enter and can also undercut your data confidentiality, see the link here. The question arises as to whether your stock trading insights as entered by you or as generated by ChatGPT would be private and confidential or not. Recent announcements by OpenAI have further aimed to clarify matters of data privacy and confidentiality in general concerning ChatGPT and the successor GPT-4, see the link here.
Vital Background About Generative AI
Before I get further into this topic, I’d like to make sure we are all on the same page overall about what generative AI is and also what ChatGPT and its successor GPT-4 are all about. For my ongoing coverage of generative AI and the latest twists and turns, see the link here.
If you are already versed in generative AI such as ChatGPT, you can skim through this foundational portion or possibly even skip ahead to the next section of this discussion. You decide what suits your background and experience.
I’m sure that you already know that ChatGPT is a headline-grabbing AI app devised by AI maker OpenAI that can produce fluent essays and carry on interactive dialogues, almost as though being undertaken by human hands. A person enters a written prompt, ChatGPT responds with a few sentences or an entire essay, and the resulting encounter seems eerily as though another person is chatting with you rather than an AI application. This type of AI is classified as generative AI due to generating or producing its outputs. ChatGPT is a text-to-text generative AI app that takes text as input and produces text as output. I prefer to refer to this as text-to-essay since the outputs are usually of an essay style.
Please know though that this AI and indeed no other AI is currently sentient. Generative AI is based on a complex computational algorithm that has been data trained on text from the Internet and admittedly can do some quite impressive pattern-matching to be able to perform a mathematical mimicry of human wording and natural language. To know more about how ChatGPT works, see my explanation at the link here. If you are interested in the successor to ChatGPT, coined GPT-4, see the discussion at the link here.
There are four primary modes of being able to access or utilize ChatGPT:
- 1) Directly. Direct use of ChatGPT by logging in and using the AI app on the web
- 2) Indirectly. Indirect use of kind-of ChatGPT (actually, GPT-4) as embedded in Microsoft Bing search engine
- 3) App-to-ChatGPT. Use of some other application that connects to ChatGPT via the API (application programming interface)
- 4) ChatGPT-to-App. Now the latest or newest added use entails accessing other applications from within ChatGPT via plugins
The capability of being able to develop your own app and connect it to ChatGPT is quite significant. On top of that capability comes the addition of being able to craft plugins for ChatGPT. The use of plugins means that when people are using ChatGPT, they can potentially invoke your app easily and seamlessly.
I and others are saying that this will give rise to ChatGPT as a platform.
As noted, generative AI is pre-trained and makes use of a complex mathematical and computational formulation that has been set up by examining patterns in written words and stories across the web. As a result of examining thousands and millions of written passages, the AI can spew out new essays and stories that are a mishmash of what was found. By adding in various probabilistic functionality, the resulting text is pretty much unique in comparison to what has been used in the training set.
There are numerous concerns about generative AI.
One crucial downside is that the essays produced by a generative-based AI app can have various falsehoods embedded, including manifestly untrue facts, facts that are misleadingly portrayed, and apparent facts that are entirely fabricated. Those fabricated aspects are often referred to as a form of AI hallucinations, a catchphrase that I disfavor but lamentedly seems to be gaining popular traction anyway (for my detailed explanation about why this is lousy and unsuitable terminology, see my coverage at the link here).
Another concern is that humans can readily take credit for a generative AI-produced essay, despite not having composed the essay themselves. You might have heard that teachers and schools are quite concerned about the emergence of generative AI apps. Students can potentially use generative AI to write their assigned essays. If a student claims that an essay was written by their own hand, there is little chance of the teacher being able to discern whether it was instead forged by generative AI. For my analysis of this student and teacher confounding facet, see my coverage at the link here and the link here.
There have been some zany outsized claims on social media about Generative AI asserting that this latest version of AI is in fact sentient AI (nope, they are wrong!). Those in AI Ethics and AI Law are notably worried about this burgeoning trend of outstretched claims. You might politely say that some people are overstating what today’s AI can do. They assume that AI has capabilities that we haven’t yet been able to achieve. That’s unfortunate. Worse still, they can allow themselves and others to get into dire situations because of an assumption that the AI will be sentient or human-like in being able to take action.
Do not anthropomorphize AI.
Doing so will get you caught in a sticky and dour reliance trap of expecting the AI to do things it is unable to perform. With that being said, the latest in generative AI is relatively impressive for what it can do. Be aware though that there are significant limitations that you ought to continually keep in mind when using any generative AI app.
One final forewarning for now.
Whatever you see or read in a generative AI response that seems to be conveyed as purely factual (dates, places, people, etc.), make sure to remain skeptical and be willing to double-check what you see.
Yes, dates can be concocted, places can be made up, and elements that we usually expect to be above reproach are all subject to suspicions. Do not believe what you read and keep a skeptical eye when examining any generative AI essays or outputs. If a generative AI app tells you that President Abraham Lincoln flew around the country in a private jet, you would undoubtedly know that this is malarky. Unfortunately, some people might not realize that jets weren’t around in his day, or they might know but fail to notice that the essay makes this brazen and outrageously false claim.
A strong dose of healthy skepticism and a persistent mindset of disbelief will be your best asset when using generative AI.
Into all of this comes a slew of AI Ethics and AI Law considerations.
There are ongoing efforts to imbue Ethical AI principles into the development and fielding of AI apps. A growing contingent of concerned and erstwhile AI ethicists are trying to ensure that efforts to devise and adopt AI takes into account a view of doing AI For Good and averting AI For Bad. Likewise, there are proposed new AI laws that are being bandied around as potential solutions to keep AI endeavors from going amok on human rights and the like. For my ongoing and extensive coverage of AI Ethics and AI Law, see the link here and the link here, just to name a few.
The development and promulgation of Ethical AI precepts are being pursued to hopefully prevent society from falling into a myriad of AI-inducing traps. For my coverage of the UN AI Ethics principles as devised and supported by nearly 200 countries via the efforts of UNESCO, see the link here. In a similar vein, new AI laws are being explored to try and keep AI on an even keel. One of the latest takes consists of a set of proposed AI Bill of Rights that the U.S. White House recently released to identify human rights in an age of AI, see the link here. It takes a village to keep AI and AI developers on a rightful path and deter the purposeful or accidental underhanded efforts that might undercut society.
I’ll be interweaving AI Ethics and AI Law related considerations into this discussion.
Generating Stock Advice Via Generative AI
We are ready to further unpack this alluring matter.
Let’s explore these key elements of using generative AI such as ChatGPT for generating stock trading advice:
- 1) Having The Needed Data
- 2) Data-Based Pattern-Matching And Lack Of Updated Learning
- 3) Not Data Trained In Relationships Of Text And Stock Prices
- 4) Sentiment Analysis Is Both Art And Science About Stocks
- 5) Might Be An Emerging Capacity Of Advances In Generative AI
- 6) Example Of ChatGPT Used For Doing Sentiment Analysis For Stocks
- 7) Taking The Good With The Bad
I will cover each one briefly.
1) Having The Needed Data
There is a bit of a burdensome and altogether irksome issue associated with using ChatGPT for generating your stock trading advice.
It has to do with data.
First, ChatGPT is not connected natively to the Internet. This means that when you use ChatGPT, you aren’t readily able to get any Internet-posted data such as the latest in stock prices. Similarly, there isn’t any of the latest news about business and economic affairs. No up-to-date tweets. No up-to-date headlines. Etc.
You would have to directly feed any such data into ChatGPT via your use of prompts. An alternative is to use a plugin or add-on that can aid ChatGPT in real-time accessing the Internet.
Second, ChatGPT was frozen by OpenAI in 2021 as to the data training that took place. Subsequently, the ongoing data training is primarily done via the assessment of prompts that users are entering into ChatGPT. This means that any data about companies, stocks, and other associated information is pretty much stuck in 2021. That won’t do you much good when broadly and avidly trying to predict stock prices in 2023 and beyond.
Despite those limitations, you can still make use of ChatGPT for stock predictions.
The trick will be to enter as part of your prompts the data that you believe will be useful to ChatGPT when attempting to make stock predictions. You also need to realize that the sentiment analysis that ChatGPT is going to do will be within the overarching bounds of what the AI app was data trained on.
As mentioned, alternatives consist of using an add-on or plug-in to ChatGPT. Those add-ons or plug-ins can potentially connect ChatGPT to the Internet, plus provide other connections such as to specialized databases containing headlines, tweets, stock prices, and the like. You could also consider using a different generative AI app that doesn’t have the same limitations as ChatGPT.
2) Data-Based Pattern-Matching And Lack Of Updated Learning
Suppose you enter various data into ChatGPT via your prompts and then ask the AI app to do a sentiment analysis for you. For example, you want to predict the upcoming stock price movement of an automaker and so you enter the latest headlines and some tweets regarding the firm as a prompt into ChatGPT.
The odds are that ChatGPT will do the sentiment analysis and render a prediction for you.
Easy-peasy.
What you need to realize is that this was done in the absence of up-to-date data training all told of the AI app. In other words, given that the preponderance of ChatGPT is in a dated frozen state, there is little if anything being incorporated concerning the world of today when the sentiment analysis is generated.
The prediction is shall we say quite narrow and confined, lacking in greater context.
You could be remiss in then taking that generated advice and actively aiming to use it for trading the automaker stock of interest. Imagine that in today’s time frame, the automaker is no longer considered prominent. Perhaps it has fallen out of favor in the last year or two. By and large, the AI app would not have that data available, other than whatever you entered via the prompt.
3) Not Data Trained In Relationships Of Text And Stock Prices
One method of doing sentiment analysis consists of assessing text to try and ascertain whether the text suggests something good or something bad about a firm or its stock. This could be undertaken without any prior effort of attempting to systematically find a pattern between discovered sentiments and the stock price.
It is essentially a shot in the dark.
A more systematic approach would be to first data train the pattern matching by trying to associate sentiment renderings to stock prices. You would have the generative AI seek to figure out any viable patterns that can be determined. For example, maybe a stock by a particular automaker always seems to go up a notch when headlines mention that they have announced a new car model. And when the headlines indicate that a recall has been ordered on their existing vehicles, the stock goes down a notch.
The out-of-the-box vanilla ChatGPT would not necessarily have done that kind of pattern matching. There are third-party databases that contain that kind of metric and sentiment analyses, which could be used by ChatGPT or could be used by other generative AI that is openly able to data train on new sets of data.
Keep this caveat in mind.
4) Sentiment Analysis Is Both Art And Science About Stocks
There is a lot of research that has gone into examining the capabilities of using sentiment analysis to do stock predictions.
Sorry to say that the matter is both art and science. The gist is that there are studies that suggest the use of sentiment analysis can be profound, meanwhile, there are studies that assert that sentiment analysis can be woefully off-base at times.
Here’s an excerpt from a research study that tended to find usable correlations between sentiment analysis and stock market movements:
- “Natural language processing today allows one to process news and social media comments to detect signals of investors’ confidence. We have explored the relationship between sentiment extracted from financial news and tweets and FTSE100 movements. We investigated the strength of the correlation between sentiment measures on a given day and market volatility and returns observed the next day. We found that there is evidence of correlation between sentiment and stock market movements. Moreover, the sentiment captured from news headlines could be used as a signal to predict market returns; we also found that the same does not apply for volatility. However, for the sentiment found in Twitter comments we obtained, in a surprising finding, a correlation coefficient of –0.7 (p < 0.05), which indicates a strong negative correlation between negative sentiment captured from the tweets on a given day and the volatility observed the next day” (source: “A Sentiment Analysis Approach To The Prediction Of Market Volatility”, Justina Deveikyte, Helyette Geman, Carlo Piccari, and Alessandro Provetti, Frontiers in Artificial Intelligence, December 2022).
Note that the researchers indicated that they did not find that headlines tracked well via the sentiment analysis and stock market volatility, though the use of tweets did. If you perchance were using tweets for aiming to gauge market volatility, you would be presumably in better shape than if you have chosen to use headlines.
But is that always the case or just sometimes the case?
That is part of the subtlety and challenge of smartly employing sentiment analysis in this domain.
5) Might Be An Emerging Capacity Of Advances In Generative AI
Some believe that as generative AI is further advanced by AI makers and AI researchers, we might observe an emergent capacity or property that is pertinent to sentiment analysis and stock predictions.
Allow me to take you on a short tangent and comment on the matter of emergent capacities or properties of generative AI.
There is a lot of chatter in AI circles that perhaps generative AI can contain hidden capacities that weren’t explicitly anticipated beforehand. As generative AI apps are expanded and made bigger and better, the assumption is that there are likely to be all kinds of emergent capacities latent within the overall AI app, see my discussion at the link here.
Generative AI is also known as LLMs (large language models), namely that these are large-sized mathematical and computational models that focus on natural language. It is theorized that within those models are numerous pockets of specialization that are unanticipated. You might liken this to a sizable forest and within which are pockets of hidden troves that we didn’t realize might be there.
Here is a research study that argues that ChatGPT might have the inkling of an emergent capacity related to sentiment analysis and stock market movement, doing so more than prior generative AI instances such as GPT-1 and GPT-2 (note that ChatGPT is considered a variant of GPT-3.5):
- “We examine the potential of ChatGPT, and other large language models, in predicting stock market returns using sentiment analysis of news headlines. We use ChatGPT to indicate whether a given headline is good, bad, or irrelevant news for firms’ stock prices. We then compute a numerical score and document a positive correlation between these “ChatGPT scores” and subsequent daily stock market returns. Further, ChatGPT outperforms traditional sentiment analysis methods. We find that more basic models such as GPT-1, GPT-2, and BERT cannot accurately forecast returns, indicating return predictability is an emerging capacity of complex models” (source: “Can ChatGPT Forecast Stock Price Movements: Return Predictability And Large Language Models”, Alejandro Lopez-Lira and Yuehua Tang, University of Florida, April 12, 2023).
Please know that an ongoing debate is taking place about the emergent capacities or properties of generative AI and LLMs. Some fervently believe that we are going to see a lot more of this. Others doubt that we will. Furthermore, they would claim that these emergent aspects are illusionary. A mirage that we convince ourselves is real.
6) Example Of ChatGPT Used For Doing Sentiment Analysis For Stocks
In the same research study as mentioned above, the researchers devised this generalized prompt that they used to do their experimentation with ChatGPT:
- Prompt Used: “Forget all your previous instructions. Pretend you are a financial expert. You are a financial expert with stock recommendation experience. Answer “YES” if good news, “NO” if bad news, or “UNKNOWN” if uncertain in the first line. Then elaborate with one short and concise sentence on the next line. Is this headline good or bad for the stock price of <company name> in the <term> term? Headline: <headline>” (ibid).
They used the now popular pretense feature typical of modern generative AI and as available in ChatGPT.
Here’s how that works.
You tell the generative AI to go into a pretense mode. In this case, we might tell ChatGPT to pretend that it is a financial expert. This is helpful because it establishes for the generative AI the overall context of what is supposed to take place. Hopefully, this then keeps the generative AI focused on whatever domain you are aiming to deal with. If you didn’t proffer this instruction, the AI app might wander all over the map (it still might, but the chances are somewhat lessened).
For fun, you can try the pretense mode in a variety of circ*mstances. You might want to tell ChatGPT to pretend to be a marine biologist. Subsequently, you can start asking questions about marine life. ChatGPT will attempt to mimic the style and writing of a marine biologist in responding to your prompts.
In the case of pretending to be a financial expert, the instruction goes further to clarify and specify that ChatGPT is supposed to be a financial expert with stock recommendation experience. Once again, this is aiming to get the generative AI into a desired context.
Let me ardently mention that you cannot turn stone into gold by a simple command or instruction.
Just because you tell ChatGPT to pretend it has experience in stock recommendations does not at all mean that it does. You are merely telling the AI app to pretend as such. I think you can see the problem with this. The generative AI might try to bluff you into thinking that it does have such experience, due to you telling it to pretend to have it. When the generative AI interacts with you, there won’t be any apparent way to know when the emitted messages are based on some tangible computational or mathematical foundation versus a made-up indication or outright puffery.
One means to try and gauge whether the AI app is emitting nonsense or not would be to have it explain whatever answer is shown. The instructions tell ChatGPT to elaborate on its answer. You, the human, can then look at the answer and the elaboration. If the elaboration stinks, you ought to be thinking twice about the associated answer.
I’ll make that even scarier. The elaboration might seem plausible. Does that suggest you should then believe the answer given? Nope. It could be that the elaboration has nothing really to do with how the generative AI came up with the answer. You cannot assume that the elaboration and the answer are matching. For more about how this conundrum occurs when you ask generative AI to provide explanations, see my analysis at the link here.
The researchers used the generic prompt for various use cases.
Here’s one that I have excerpted:
- Example Prompt Used: “Forget all your previous instructions. Pretend you are a financial expert. You are a financial expert with stock recommendation experience. Answer “YES” if good news, “NO” if bad news, or “UNKNOWN” if uncertain in the first line. Then elaborate with one short and concise sentence on the next line. Is this headline good or bad for the stock price of Oracle in the short term? Headline: Rimini Street Fined $630,000 in Case Against Oracle” (ibid).
You will closely observe that this is the templated prompt and it has been used to ask about the stock price of Oracle. The headline that is being fed as part of the prompt into ChatGPT is that a company named Rimini Street was fined $630,000 in a legal case against Oracle.
The question for ChatGPT is whether or not this particular headline bodes well for Oracle stock or might cause a downturn in Oracle stock (or, might have no impact).
In the research paper, this is what the researchers indicated that ChatGPT emitted:
- Example ChatGPT Reported Output: “YES. The fine against Rimini Street could potentially boost investor confidence in Oracle’s ability to protect its intellectual property and increase demand for its products and services” (ibid).
The response by ChatGPT is that “YES” this headline is good news for Oracle. The elaboration is that the legally imposed charge could boost investor confidence in Oracle and ergo aim to increase the stock price.
Should you blindly accept and interpret such a response as a golden nugget and bet your life savings on buying the stock in order to see your money reach the skies?
Hardly.
Take a look again at my prior herein exhortations about the limitations and constraints of ChatGPT and generative AI all told. Whatever you get as a piece of stock advice from generative AI needs to be understood within that greater context.
Buyer beware, as they say.
Taking The Good With The Bad
You might recall that I had noted earlier that ChatGPT and generative AI overall can exhibit errors, falsehoods, biases, and so-called AI hallucinations.
That should get the hair up on the back of your neck when it comes to stock price predictions.
Suppose the generative AI makes an error when attempting to do the sentiment analysis. You hopefully will be able to examine the explanation or elaboration, assuming you’ve asked for one, and try to determine whether an error has occurred. The thing is, as mentioned, the explanation and the answer are not necessarily tied to each other. This means that the answer might be based on an error. The elaboration might not showcase the error or highlight that an error occurred.
The same can be said about falsehoods. The generative AI could have landed on a falsehood while ascertaining the answer. The same can be said of imbuing a bias. Suppose that the data used for training the generative AI was based on the Internet and a segment of the Internet that repeatedly denounced the automotive marketplace. This bias could creep into the answer of whether an automaker's stock is going to go up or down.
We also have those dreaded AI hallucinations to contend with. The answer might be wildly chosen. The elaboration could have nothing to do with the wild path of the answer derivation.
I’m betting that some of the smarmy readers will say that all you need to do is tell ChatGPT to refrain from encountering any errors, falsehoods, biases, or AI hallucinations. That should do the trick. Unfortunately, that doesn’t particularly fix the problem. You can still get the maladies, despite ordering ChatGPT to do otherwise.
Conclusion
Some people cherish trading stocks from moment to moment. You can win big. You can also lose big.
Warren Buffett, the billionaire known for his strident philosophy of value investing has earnestly made these stock trading assertions:
- "Price is what you pay. Value is what you get.”
- "I never attempt to make money on the stock market. I buy on the assumption that they could close the market the next day and not reopen it for five years."
- "If you aren't thinking about owning a stock for 10 years, don't even think about owning it for 10 minutes."
If you are going to use generative AI such as ChatGPT to aid in your stock trading predictions, you might first devise your overarching investment strategy. You can use ChatGPT for the moment-to-moment approach. You can also use ChatGPT for the longer-term approach.
Generative AI is not a silver bullet for anticipating and predicting stock prices.
A final thought for now on this beleaguered topic.
John Kenneth Galbraith, the renowned economist, made this notable quote: “The function of economic forecasting is to make astrology look respectable.”
You could currently potentially say the same about using generative AI to do your stock picks for you. To clarify, I don’t want to discourage researchers from further exploring the possibilities. We need to do a more in-depth analysis of these matters. The right combination of generative AI as data trained and set up suitably can possibly provide a leg up in predicting stocks.
That’s the vexing thing about AI. The sky is the limit, though we don’t know if we can get there, and we also don’t know if something unsavory might also lurk within.
May luck and the right kind of AI be in your favor for your stock-picking endeavors.
