Boosted.ai CEO and co-founder Josh Pantony joined “Mornings with Maria” to weigh fears of AI-driven market disruption, the future of white-collar work, and the limits of machines in stock picking. The conversation, aired from New York, probed how fast automation is moving, which jobs may shift, and whether investors can trust algorithms during volatile periods.
Where AI Fits in Finance Today
Financial firms have used algorithms for decades, but the latest wave of tools can read filings, scour earnings calls, and process news at high speed. Pantony described how investors now blend natural language models with traditional factors to screen risks and find themes across sectors. These systems rank companies, surface unusual signals, and flag changes in sentiment faster than human teams can.
Even so, data quality remains a hurdle. Models can latch onto noisy signals or overfit to brief windows. Many funds now maintain guardrails: clear data pipelines, version control, and stress tests across cycles. The goal is not a magic stock picker, but a tool that adds discipline and speed to a human process.
White-Collar Work Under Pressure—or Evolving
Concerns about job loss came up often. Entry-level research tasks, first drafts of notes, and basic screening are prime targets for automation. That could shrink some roles, especially in research support and operations. But Pantony emphasized new needs as well: prompt design, data vetting, and model risk oversight.
Firms that keep headcounts steady are shifting duties. Analysts spend less time cleaning data and more time testing ideas, meeting management teams, and communicating risk. Productivity gains favor teams that pair domain knowledge with careful model governance.
Can AI Predict the Stock Market?
“Can artificial intelligence truly predict the stock market?”
That question framed the heart of the debate. Academic work on efficient markets suggests prices already reflect most public information. AI can help identify short-lived signals, hidden patterns in text, and cross-asset relationships. But it cannot see the future or rewrite market math.
Pantony drew a line between prediction and probability. Well-built models can tilt odds, not guarantee outcomes. They are strongest when combining diverse inputs—fundamentals, alternative data, and macro clues—and when users track error rates. Performance should be judged across regimes, not just during quiet rallies.
Risk, Regulation, and Trust
Model risk is a growing focus for compliance teams. Financial regulators have asked firms to document data sources, test for bias, and explain decisions that affect clients. Black-box systems raise concerns when trades move real money without clear reasoning.
Best practice now includes human sign-off, scenario analysis, and clear audit trails. Pantony noted that explainability wins trust with boards and clients. If a model recommends a trade, teams should know which signals drove the call and how sensitive it is to shocks.
What to Watch Next
Several trends will shape the next year. Costs for high-quality data and computing are rising, which may widen the gap between large asset managers and smaller firms. At the same time, open-source tools are improving fast, lowering barriers for niche strategies.
- Hybrid teams that pair quants with sector experts are gaining an edge.
- Text and audio analysis of earnings calls is moving from novelty to standard input.
- Clients are asking for clearer explanations of AI-driven decisions.
For investors, the key is discipline. Treat AI as a decision aid with measured expectations. Track drawdowns, not just headline returns. Demand transparency on data and methods. And remember that liquidity, costs, and risk controls matter as much as any signal.
Pantony’s message was pragmatic: automation is changing workflows, not replacing judgment. Firms that set guardrails, test models across cycles, and keep experts in the loop are more likely to benefit. The next phase will reward teams that prove their tools work in rough markets, not just calm ones.
