The Prompt Paradox

by Craig Basinger, Purpose Investments, David Benedet, Portfolio Manager, Purpose Investments

Artificial intelligence (AI) has clearly been a dominant market, economic and workflow theme over the past few years following the proliferation of large language models (LLMs). Many have started to use some AI tools in their day-to-day; many have not. Some believe it will change our world; some believe it is simply another technology tool to help us be more productive, perhaps with a financial bubble along the way. This ethos is not about how much compute we need and how we will power those data centres. It is about using AI.

We are not super users of AI, but we do use it across various areas. Years before these LLMs hit the public’s consciousness, we started dabbling in machine learning related to our more quantitative research and management. Now, with the continued advancement of LLMs, the use cases of AI have really increased.

Summarization is a big one. AI-generated summaries of earnings calls are very useful and time-saving, with limited loss of content. Report summaries are also useful. Those great 100+ page analyst reports are the best, but sometimes reducing them can enable us to get through more of them. For the record, we still read some from cover to cover (in case any of the analyst authors of such reports are reading the ethos).

Upload a report or reports into ChatGPT and ask it to summarize them. Or better yet, ask it more specific questions like, “What is the biggest concern of management? Or risk to margins?” By extracting information across various sources, it can uncover things that may not have been on your radar.

Enrichment is quickly becoming a valuable output from LLMs. Have you ever received a generic email trying to market or sell you something? Maybe even from an asset management firm with a really great fund or ETF? The low open rate on such communications leaves much to be desired. Using AI to create much more customized communications, incorporating specifics about the client/recipient, can greatly improve the readership. Even using tools to make communication content can save time and improve quality.

Search is quickly becoming more user-friendly with LLMs. Everyone is familiar with a traditional search on a given topic, such as Google. You receive a list of potentially applicable links to content on your inquiry. Sure, you have to scroll past the ‘sponsored’ links companies have paid to show up on the topic and then spend time sifting through the rest to find what you are looking for. LLMs such as ChatGPT certainly provide a much more succinct output that is often much more on point. We use search engines much less now and encourage readers to try it out on LLMs, if you have not already.

Many tasks such as summarizing market news, analyzing data, or drafting client communications can now be prepared using LLMs. Even the title of this report was generated using an LLM. Pretty good, right? While it is true that AI will allow us to save time by making information more accessible, relying too much on it might weaken our ability to think critically. Mainly because it is too easy and too compelling.

Traditional search engines are more time-consuming, but you also enjoy better clarity on the source. For instance, search for energy demand in 2025 and you see a number of links. Some are from very reputable organizations such as the International Energy Agency (IEA) or British Petroleum. Other links are from sources, let’s say, that are more suspect. Like a blogger in their basement (exaggerating for effect). The reader can decide which sources to put more credence behind or read all and use that important critical thinking skill to come to their own conclusion.

When you ask an LLM a similar question, the output is on point, well-constructed and easy to read. There may be sources cited, but it’s challenging to differentiate between content from a more trusted source or one that you may view as less objective, as the output is an amalgamation of many sources.

Prompt Paradox – The answer you receive from an LLM is based on both a slice of available information and on the prompts you feed into it. How you ask or frame the question impacts the answer or output, which means the LLM can just as easily reinforce existing biases as it can uncover new insights. If you were to word the prompt in a more bullish tone, you’d likely get an associated output you were hoping for, not necessarily the right one or one that would encourage critical thinking. Learning to use LLMs as a helper rather than a substitute for thinking will be vital. Otherwise, you’re just getting a high-tech “Yes-man”.

The following is an oversimplification and exaggeration, asking a popular LLM why someone should buy or sell gold.

Prompt: Why should I buy gold?

Prompt: Why should I sell gold?

Gold remains a popular investment due to its role as a safe haven in times of economic uncertainty and its ability to diversify a portfolio. Here’s why it might be a good addition to your investment strategy:

Hedge Against Inflation:
Gold often holds its value when paper currencies decline, offering protection against inflation and currency risk.

Portfolio Diversification:
Its performance typically differs from stocks and bonds, which can reduce overall portfolio volatility.

Safe Haven Asset:
In periods of market or geopolitical turbulence, investors often turn to gold, considering it a stable asset when other investments face stress.

Liquidity and Global Demand:
Gold is highly liquid and is in demand worldwide, adding to its attractiveness as a long-term store of value.

Limited Supply:
Being a finite resource, gold’s scarcity may help support its price over the long run.

While gold can be volatile in the short term and does not provide income like dividends or interest, it may still be a wise choice for preserving wealth and mitigating overall risk.

Selling gold can be a strategic move for several reasons:

Liquidity:
Gold is a tangible asset that can be quickly converted to cash, which is useful for emergencies or new opportunities. Capturing gains when prices are high may also be advantageous.

Market Conditions:
Gold prices often peak during periods of economic uncertainty or geopolitical tensions. Selling when demand is high can secure profits before any potential downturn.

Portfolio Management:
If gold has grown to dominate your portfolio, selling some can help rebalance and reduce concentration risk. Converting gold to cash also provides capital for diversifying into other asset classes that may offer better returns or income, like stocks or bonds.

Opportunity Cost:
Since gold doesn’t generate income like dividends or interest, reallocating funds could allow you to invest in assets that yield regular returns.

Ultimately, the decision to sell should align with your financial goals and market outlook. Consulting a financial advisor can help ensure your decision fits your overall strategy.

One output is rather positive on gold, while the other is rather cautious. While this comparison in the prompt is blatant, the challenge comes when even minor changes in the wording of a prompt have a significant impact on the response. The user may not be aware or even understand why changing a few words results in a different output. The difference in the verdict is not only determined by the info provided in the prompt, but it highlights how much results can depend on the written prompt – just ask the newest professional role in AI, the prompt engineer. The ability to precisely articulate what you're looking for in a prompt is critical.

The wording in your prompts can fuel confirmation bias and, since it is rather opaque how an LLM moves from prompt to response, it may be hard to detect. Too much confirmation bias hurts critical thinking.

Remember in school, you were asked to write a book report or a report on a specific topic? Nobody liked it – write 2,000 words on topic X. And now people have LLMs to easily create for them. But what might not have been apparent at the time of doing those book reports was that the task was actually designed to foster critical thinking. You were forced to research, determine what is important and what is not important, and formulate your report.

One of the risks of AI and LLMs is a further deterioration of people thinking critically about topics. Counter this is the increased ease of gathering information on a topic. You just have to be careful how you ask.

Final Thoughts

The benefits of AI are very impressive and we are still in the early days of application development. Sure, there are risks such as higher confirmation bias, but the benefits are so much greater. Regardless of your choice of application, explore. It clearly adds to productivity.

For fun, we put our content through our preferred AI detector. It uses AI to detect AI. 12% of text appears to be AI-generated, roughly the same amount of content in the side-by-side comparison. The rest is original content, in case you were getting worried.

— Craig Basinger is the Chief Market Strategist at Purpose Investments

— Brett Gustafson is an Associate Portfolio Manager at Purpose Investments

Get the latest market insights in your inbox every week.

Sources: Charts are sourced to Bloomberg L. P.

The content of this document is for informational purposes only and is not being provided in the context of an offering of any securities described herein, nor is it a recommendation or solicitation to buy, hold or sell any security. The information is not investment advice, nor is it tailored to the needs or circumstances of any investor. Information contained in this document is not, and under no circumstances is it to be construed as, an offering memorandum, prospectus, advertisement or public offering of securities. No securities commission or similar regulatory authority has reviewed this document, and any representation to the contrary is an offence. Information contained in this document is believed to be accurate and reliable; however, we cannot guarantee that it is complete or current at all times. The information provided is subject to change without notice.

Commissions, trailing commissions, management fees and expenses all may be associated with investment funds. Please read the prospectus before investing. If the securities are purchased or sold on a stock exchange, you may pay more or receive less than the current net asset value. Investment funds are not guaranteed, their values change frequently, and past performance may not be repeated. Certain statements in this document are forward-looking. Forward-looking statements (“FLS”) are statements that are predictive in nature, depend on or refer to future events or conditions, or that include words such as “may,” “will,” “should,” “could,” “expect,” “anticipate,” intend,” “plan,” “believe,” “estimate” or other similar expressions. Statements that look forward in time or include anything other than historical information are subject to risks and uncertainties, and actual results, actions or events could differ materially from those set forth in the FLS. FLS are not guarantees of future performance and are, by their nature, based on numerous assumptions. Although the FLS contained in this document are based upon what Purpose Investments and the portfolio manager believe to be reasonable assumptions, Purpose Investments and the portfolio manager cannot assure that actual results will be consistent with these FLS. The reader is cautioned to consider the FLS carefully and not to place undue reliance on the FLS. Unless required by applicable law, it is not undertaken, and specifically disclaimed, that there is any intention or obligation to update or revise FLS, whether as a result of new information, future events or otherwise.