AI vs. analysts: The future of investment research

Can artificial intelligence make human analysts irrelevant? That’s the question on everyone’s mind as AI models completely revolutionize investment research. Byron Wien, a market strategist who defined the 1990s, believes the best research comes from bold, non-consensus ideas that prove correct.

Now the pressure is on AI to meet this standard and potentially sideline analysts who have dominated the field for decades. For years, analysts have dissected financial statements and scoured headlines, all to help investors make better decisions.

AI has stepped into this space with tools that simplify, automate, and sometimes outperform traditional methods. Large language models (LLMs) have become particularly effective at analyzing financial data, doing in minutes what might take a team of analysts days.

Predicting earnings, for instance, plays right into AI’s strengths. Profit patterns tend to follow logical trends—good years lead to more good years; bad years lead to more bad ones. AI thrives in these predictable spaces, outperforming human analysts who sometimes let noise or bias cloud their judgment.

LLMs rewriting the investment analysis playbook

The University of Chicago’s work with LLMs has turned heads. Researchers used AI to predict earnings variance and found that these models beat human analysts’ median estimates. The secret? LLMs excel at understanding the story behind earnings reports, something traditional algorithms never managed to do. 

These models mimic the logical steps of senior analysts, like disciplined juniors on a financial team. AI models also sidestep one of the biggest human pitfalls: overconfidence. Analysts are notorious for adjusting their projections to fit what they think investors want to hear. AI doesn’t play that game.

By tweaking an AI model’s “temperature” settings—a fancy term for randomness—you can calculate risk and return bands with cold, hard statistics. You can even get a confidence estimate for its predictions. Humans, by comparison, tend to get cocky with their forecasts, doubling down on bad calls instead of reassessing.

Despite these wins, AI is far from perfect. It won’t find the next Nvidia or foresee another global financial meltdown. Big market shocks like these don’t follow patterns, and AI struggles when the unexpected happens.

It also can’t grill company executives during earnings calls or pick up on evasive answers about critical issues. Markets are messy and constantly shifting, and AI lacks the intuition to adapt. That’s where top analysts still shine—they know when to pivot, dig deeper, and push for answers.

But the AI hype will probably remain strong for a long time. Tech giants are obsessed. Microsoft is betting big—$80 billion big—on AI and the infrastructure it needs. For fiscal 2025, the tech giant plans to spend more than half of that in the U.S. on data centers to train and deploy AI models.

Why the splurge? AI demands insane computing power. Training models like ChatGPT means linking thousands of chips in massive data center clusters.

Advertising dollars could power the next tech boom

AI might follow the same road as past tech revolutions: fueled by advertising money. Remember how Google and Facebook rose to power? They cashed in on brand-building budgets, taking dollars from everyone—from Tide to your local plumber.

Even subscription-heavy companies like Netflix and Amazon are now leaning on ads. Alphabet, Google’s parent company, is a prime example of how far this model can go. Since its 2004 IPO, Alphabet’s revenue has surged by 160 times, hitting over $300 billion in 2023.

AI has the potential to reshape industries, just like radio, TV, and the internet did before. Back in the day, newspapers relied on ads for two-thirds of their revenue.

Radio and TV thrived on commercials, keeping them free for audiences. AI might soon be the next big advertising platform, pulling in dollars to fund groundbreaking developments.

AI can spit out ideas—some brilliant, some nonsensical. It can run endless scenarios, pulling insights from history that even an army of researchers might miss. But it can’t give you that “spark of genius.” Analysts bring something AI can’t replicate: the ability to question, adapt, and see the bigger picture in real time. 

That human touch is still invaluable in a world where non-consensus recommendations—the ones no machine would think to make—often turn out to be the most profitable. The real takeaway? AI and analysts aren’t enemies. They’re tools for each other.

A Step-By-Step System To Launching Your Web3 Career and Landing High-Paying Crypto Jobs in 90 Days.