Will Employing AI Instead of Humans Really Help Companies’ Bottom Lines?

Key Takeaways

• MIT study: AI can automate 11.7% of tasks, $1.2 trillion wages
• AI tool spending reached $37 billion in 2025, still low‑priced
• Pricing may rise as providers shift to usage‑based models
• Customer‑support chatbots can handle volume cheaper than hiring staff
• HR leaders will lock in contracts, scrutinize AI ROI

Pulse Analysis

The promise of generative AI has spurred multi‑billion‑dollar bets from firms such as OpenAI and Anthropic, which together burned over $10 billion in 2024 without profit. Companies poured an estimated $37 billion into AI stacks in 2025, lured by subscription prices often lower than a single employee’s cost. Analysts warn the “free‑trial era” is ending; as infrastructure costs rise, providers are testing usage‑based pricing and higher seat fees. This shift threatens the cost advantage that initially justified large‑scale AI adoption.

AI’s impact is best measured in tasks, not whole jobs. An MIT study from late 2025 found AI can automate about 11.7 % of work tasks, equivalent to $1.2 trillion in annual wages. Gains concentrate in repetitive, high‑volume functions such as first‑tier customer support, where chatbots handle thousands of inquiries without extra headcount. In analytics, Visier’s V speeds query responses for a $100 k analyst, but human oversight remains essential for framing questions and delivering insights. Virtual teaching assistants similarly augment student access without displacing faculty.

Rising prices are already reshaping procurement. HR leaders shift from open experimentation to one‑year lock‑in contracts to secure predictable OPEX and avoid token spikes. Companies must now consider total cost of ownership, including engineering talent for integration and maintenance. As pricing moves toward utility‑style billing, firms that demonstrate clear ROI—through reduced labor spend, faster time‑to‑market, or higher quality output—will keep AI as a strategic asset, while others may revert to hybrid or open‑source alternatives.