The 4 Components of AI Investing and How ChatGPT Falls Short

ChatGPT is trending right now. Since launching on November 30, 2022, the OpenAI technology that writes everything from term papers to song lyrics has been criticized as dangerous and shallow1 and banned from public school systems2. At the same time, others are heralding it as the next generation of AI with the potential to upend the white-collar workforce3. But when ChatGPT is broken down to its basic components, we realize its promise may be exaggerated. From an investment standpoint, it lacks across the four key areas necessary for smart AI investing: data, algorithms, applications, and processing power.

Data is the most important component of AI and Deep Learning (AI/DL). Without it, algorithms can’t exist, applications are limited, and processing power is irrelevant. ChatGPT4 launched with a very limited batch of data. It is not connected to the internet, occasionally produces incorrect answers, and has a limited knowledge of world events after 2021. It is learning as more people try it, but it can only ever be as good as its inherently fallible users.

Algorithms rely on the quality and quantity of data for high performance. The same algorithm becomes even more powerful with the more data it ingests. The more shows we watch and rate on Netflix, the better recommendations we get for future viewing. But not all algorithms offer the same level of sophistication. In the words of the OpenAI CEO2, “ChatGPT is incredibly limited, but good enough at some things to create a misleading impression of greatness.”

With more high-quality data and sophisticated algorithms, AI/DL can be used for a wider variety of applications than ever before. Companies that expand their technologies to new and diversified applications are those poised for growth and less prone to downside risk. While inflated as nothing short of world-changing, ChatGPT is merely a language model with limited, albeit groundbreaking, applications. Plus, many of the applications it is intended for are pushing against it as a threat to learning and creativity.

The more applications to which an AI/DL algorithm can be applied, the more important it is that the processing power keeps up. Processing power can be thought of as a technology’s capacity to manipulate data with efficiency, accuracy, and speed. With all the attention ChatGPT is receiving right now, its processing power has been tested and has already failed. This screenshot really says it all.

At TrueShares, the philosophy behind our Technology, AI and Deep Learning ETF (LRNZ) has always been to provide exposure to modern asset classes through active management of secular growth companies that provide behind-the-scenes technologies across industries. 

For data exposure, LRNZ holds companies like Snowflake which deals in the business of data mobilization for all industries of all scales. For algorithm exposure, Nvidia is one of many LRNZ holdings with an entire division dedicated to developing new algorithmic techniques with advanced and varied capabilities and efficiency. For wide application exposure, LRNZ holdings are highly diversified, from biotech holdings like AbCellera, Prime Medicine, Schroedinger, and Relay Therapeutics to cybersecurity companies like Sentinel One. For processing power exposure, AMD develops high-performance computer processors used in myriad ways by nearly every industry from education to automotive. For a list of the current fund holdings, please visit

The 20 to 30 concentrated positions in LRNZ are designed to provide exposure to Artificial Intelligence and Deep Learning and to benefit from the sector’s potential. We don’t chase trends or speculation. We dig into the fundamentals of every holding. We even had a real human write this blog so you could learn more about it.