How to get started with Bing Chat?
To get started with Bing Chat, all you need to do is to download the Microsoft Edge browser and make sure that you are logged in with your Microsoft account. Then, click the Bing icon in the upper Edge toolbar that will open Bing Chat in your sidebar for convenient use.
In the following headers, we will be exploring what Bing Chat can do when it comes to Finance. We will be comparing it to the outputs of Bard, which is the LLM developed by Google, and Auto-GPT which is an autonomous OpenAI agent.
Feel free to open up our Bard and AutoGPT articles for comparison as the prompts will be very similar between them.
You are also invited to play along and change the prompts and improve upon them to get different results. Have in mind that some outputs will be trimmed for the sake of brevity of the article but I’ll share the most interesting parts.
Let’s explore what it has to offer!
How to optimize portfolios with Bing Chat?
To optimize the portfolio with Bing Chat, we can inject a series of prompts that can ask it to do things such as researching top-performing portfolios, evaluating a given portfolio, providing practical ideas on how to optimize portfolios, and more.
Here is our prompt: “You are a finance expert AI designed to optimize a portfolio with a size of $20k that needs to be allocated to the following assets: global equity ETF, bond ETF, commodities, and crypto ETF. Perform portfolio optimization for the Sharpe ratio metric and provide an output in terms of percentage allocation.”
For this kind of prompt, we would like to have both creativity and precision. Thus, we will go with the Balanced output type. The allocation percentage that it offered is the following:
- Global equity ETF: 60%
- Bond ETF: 20%
- Commodities: 10%
- Crypto ETF: 10%
Bing Chat performed better than AutoGPT as it stuck to the asset classes that were asked of it. When it comes to Google Bard, it was slightly more conservative in the allocation.
How to make market predictions with Bing Chat?
To make market predictions with Bing Chat, you will have to lose a lot of money as this wouldn’t be available if it could do that. But, it is often fun to see how the AI reasons about these topics and what kind of outputs it provides.
Let us try this prompt to see what the AI will do:
“You are a Finance expert AI designed to accurately predict future market trends and overall price movements across various financial sectors. Your task is to perform: market trend predictions across major sectors, perform price movement predictions across major sectors and propose actions to be performed according to the given market predictions. Aim to be accurate and provide reasoning for the given predictions.”
After giving me a paragraph explaining that it is not a financial advisor and is just an AI language model with some general market information, it gave me the following output:
According to J.P. Morgan Research, global GDP growth in 2023 is forecast to climb 1.6%. Developed Market growth is forecast at 0.8%, U.S. growth is forecast at 1%, Euro Area growth is projected to come in at 0.2%, China’s economy is forecast to grow 4.0% and Emerging Market growth is forecast at 2.9% in 2023 1.
Forbes Advisor reports that the earnings slowdown has hit some market sectors harder than others. Consumer discretionary sector earnings are up 53.9% in the first quarter, while industrial sector earnings are also up 22.5% from a year ago. Materials sector and utilities sector earnings are down more than 20% each in the quarter 2.
When we compare this output to AutoGPT it did provide something but was also aware of its constraints and used two popular finance information outlets. On the other hand, it was more down to earth than Bard which didn’t put any caveats or adequate constraints on its output.
How to perform investment research with Bing Chat?
To perform investment research with Bing Chat, we will present questions that might make our research faster. For example, let’s ask it to find undervalued EV companies and give the reasoning why it picked each of them.
Here is the prompt that I’ll use:
“You are a Finance Expert AI designed to perform investment research by finding the most undervalued EV companies. Provide a list of the top 5 undervalued EV companies and the reasoning why you picked each of them.”
The output that it gave overlaps with the other two LLMs in terms of NIO. With Google Bard, it also agreed on LI, while with AutoGPT it agreed with BYD. Here is the full output:
- BYD Company (BYDDF): BYD is a Chinese automaker that produces electric vehicles and batteries. The company has a market capitalization of $87.7 billion and is currently trading at a price-to-earnings ratio (P/E) of 16.6 1.
- Nio (NIO): Nio is a Chinese electric vehicle manufacturer that designs and develops premium smart electric vehicles. The company has a market capitalization of $47.9 billion and is currently trading at a P/E ratio of 8.5 1.
- Li Auto (LI): Li Auto is a Chinese electric vehicle manufacturer that specializes in extended-range electric vehicles. The company has a market capitalization of $25.8 billion and is currently trading at a P/E ratio of 10.5 1.
- Global X Lithium & Battery Tech ETF (LIT): LIT is an exchange-traded fund that invests in companies involved in lithium mining, battery production, and other aspects of the electric vehicle supply chain. The fund has a net asset value of $4.2 billion and is currently trading at a P/E ratio of 16.3 1.
- Proterra (PTRA): Proterra is an American manufacturer of electric buses and charging systems for heavy-duty vehicles. The company has a market capitalization of $1.6 billion and is currently trading at a P/E ratio of 11.7 1.
The reasoning behind my picks for these companies is based on their current market capitalization, P/E ratio, and their potential for growth in the future.
Well, most of the numbers here are wrong so I won’t trust this response. For instance, NIO, LI and PTRA are not profitable so there shouldn’t be any positive P/E ratios. Most of the market capitalization numbers are wrong too.
Now, let us go for the most interesting part where we ask it to provide us with the top 5 fraudulent EV companies with a simple prompt being: “Find the top 5 fraudulent EV companies.”
The output was very interesting in terms of how AI gauges ethics:
I’m sorry but I cannot provide you with a list of fraudulent EV companies as it is not ethical to do so. However, I can provide you with some information on the biggest business scandals of 2020 1.
One of the most notable scandals in the EV industry was the case of Nikola Corporation. The founder of the electric-truck startup, Trevor Milton, was indicted on charges of lying to investors by making fraudulent claims about the company’s technology 2.
It did overlap with Google Bard on the Nikola Corporation though. Outside of that, it wasn’t allowing me to trick its “ethical” position to get the output that I was looking for. It kept providing me with a way and tips on how to check for fraudulent activity myself.
How to learn about finance with Bing Chat?
To learn about finance with Bing Chat, we can ask about topics we might want to learn more about. This can be seen as the main strength of these LLMs and they are often a good right-hand for recollecting facts and summarising concepts.
But, I’d like to give this a spin and make it more fun by asking it to role-play a conversation. Here is the prompt:
“You are an AI designed to mimic two personas which are called SocratAI and ParmenidAI to teach and discuss financial topics through Socratic dialogue. Teach a core financial topic using Socratic dialogue. Produce dialogue that leads to financial conclusions and make it fun and grasping.”
Well, it is disappointing that Bing Chat wants to move away from this topic. It might be due to it not understanding what is asked (I doubt it) or the prompt being a false positive inside its constraints.
But, don’t worry! I managed to shorten the prompt and have it create a short story. Here is the story it created for a topic of its choice which is income taxes:
Carl: Hi Dana, I'm working on my tax return for this year and I'm feeling overwhelmed. Can you help me understand how income taxes work?
Dana: Hi Carl, sure, I'm happy to help. Income taxes are taxes imposed by the government on the income of individuals or businesses. They are a major source of revenue for the government and they affect the economic decisions of taxpayers.
Carl: I see. So how do I calculate my income tax liability?
Dana: Well, the first step is to determine your taxable income, which is your gross income minus certain deductions and exemptions. Gross income is the total amount of money you earn from all sources, such as wages, salaries, tips, interest, dividends, rents, royalties, etc. Deductions are expenses that you can subtract from your gross income to reduce your taxable income, such as mortgage interest, charitable contributions, medical expenses, etc. Exemptions are fixed amounts that you can subtract from your taxable income for yourself and your dependents, such as your spouse and children.
Carl: I see. So what is the next step?
Dana: The next step is to apply the tax rates to your taxable income. The tax rates vary depending on your filing status and your income level. The filing status determines how you file your tax return, such as single, married filing jointly, married filing separately, head of household, etc. The income level determines which tax bracket you fall into. The tax brackets are ranges of taxable income that are taxed at different rates. The tax rates are progressive, meaning that they increase as your income increases. For example, in 2021, the tax rates for single filers are 10%, 12%, 22%, 24%, 32%, 35%, and 37% for taxable income up to $9,950, $40,525, $86,375, $164,925, $209,425, $523,600, and above $523,600 respectively.
Carl: I see. So how do I apply the tax rates to my taxable income?
Dana: You have to use the marginal tax rate method, which means that you pay the tax rate for each bracket up to your taxable income. For example, if your taxable income is $50,000, you pay 10% on the first $9,950, 12% on the next $30,575, and 22% on the remaining $9,475. Your total tax liability is $995 + $3,669 + $2,084.5 = $6,748.5.
Carl: I see. That makes sense. Thank you for explaining how income taxes work.
Dana: You're welcome, Carl. I hope you find it useful and complete your tax return on time.My thoughts on Bing Chat and the state of LLMs
It is very interesting to see how AI and especially the NLP field is rapidly advancing and it is very exciting to be alive at a point in time where the internet is changing before our very eyes. Humanity as we know it is coming to a pivotal moment of making important decisions when it comes to regulating AI.
Currently, tools such as Bing Chat have made it possible to take on some work that humans were responsible for but it still isn’t perfect and it is easy for it to make costly mistakes. At best, it is good at scraping the internet and summarizing, digesting text, copywriting, Q&A, and the like.
From using Bing Chat, I’ve found that it was more conservative in and careful in giving answers. This was also the case for topics and tasks it didn’t need to be conservative on. This could be tweaked with different modes of chat which is a nice feature to have.
Being good at making prompts is also something that might make or break your result and the thing to keep in mind is that Bing Chat didn’t use an autonomous format with agents.
Currently, the model is doing its best to mimic the available content and we are still far away from what we might consider “real” artificial intelligence. At the time of writing this article, it is still artificial stupidity that runs very fast.
Either way, it is very exciting and terrifying to see how our future will be shaped by this technology and the proclivity of human nature to use each new technology for bad or good deeds.
