Legacy companies will struggle with AI

Legacy companies are slow to roll out AI tools due to risk aversion and bad incentives

Yesterday, I came across an intriguing LinkedIn post by Ted Merz, a longtime Bloomberg News executive turned entrepreneur. Merz discussed SigTech, a financial analytics startup launching a no-code solution designed to democratize access for firms and analysts with limited programming resources. He highlighted that this innovation is part of a broader transformation on Wall Street, where natural language processing (NLP) and generative AI are delivering solutions previously deemed unimaginable.

Merz illustrates how SigTech’s software can interpret natural language instructions and perform complex analyses. For example, he provides this prompt: “Construct a rolling futures strategy for the past two years for gold, silve, and Bitcoin, allocating 15%, 35%, and 50% to each. Analyze and chart its performance.” This capability is a game-changer for Wall Street, and Merz’s post offers further insights into its potential impacts.

However, I’d like to shift the focus to a related issue that Merz briefly touches on—the adoption of AI by legacy companies. Merz states, “The dominant platforms in the financial information business—Bloomberg, FactSet, and Refinitiv Eikon—have been slow to roll out AI features.” I responded to his post, noting that this doesn’t surprise me. Last year, Bloomberg generated buzz with BloombergGPT, an attempt to emulate OpenAI’s ChatGPT success. However, updates on BloombergGPT have since been scarce, likely for good reason.

The core issue for Bloomberg and other legacy companies capable of developing AI features is their inherent organizational risk aversion. Executives at these companies, shaped by their training and professional demeanor, tend to avoid risky projects. Consequently, unreliable and potentially hallucinatory AI technologies are unlikely to be pursued, despite these technologies posing existential threats to legacy revenue streams.

This dynamic isn’t intuitive to many technologists, especially those who flourish in startup environments. It’s not just financial analytics firms that grapple with this issue; even tech giants like Google face similar challenges. Google’s hesitance to release AI products stems partly from concerns over model bias, but primarily from the existential threat AI poses to its ad-based search business. When a company’s revenue heavily relies on a particular model, executives are incentivized to protect it at all costs, even if a new technology threatens those legacy revenue streams.

Legacy firms’ risk aversion creates a significant gap between innovative startups and established industry giants. This gap presents a unique opportunity for entrepreneurial ventures to step in and provide cutting-edge AI solutions tailored to specific industry needs. Companies like SigTech in financial analytics, or John Snow Labs in pharma research, exemplify how startups can thrive by offering specialized AI capabilities that legacy firms are hesitant to develop internally.