Substrate
ai

Studies Question ROI for AI Agents

Multiple research efforts have found limited financial returns from generative AI deployments in business settings. A similar pattern appears to be emerging for AI agents according to one analysis. The findings contrast with widespread industry optimism about the technology.

GA
nypost.com
yourtango.com
3 sources·May 8, 2:52 PM(14 hrs ago)·1m read
Studies Question ROI for AI Agentsyourtango.com
Audio version
Tap play to generate a narrated version.

Multiple academic studies have concluded that generative AI has delivered limited return on investment for most businesses. One study from MIT reached that conclusion along with six or seven others from separate research teams that produced similar results.

The pattern for AI agents appears comparable with substantial promotional attention but limited measured financial benefits to date. The observations come from independent analyses rather than industry surveys that often accompany product launches. Business adoption of generative AI tools has expanded rapidly since the technology gained public attention in 2022.

Companies have invested in applications ranging from content creation to customer service automation. Returns have varied by sector and implementation approach according to the body of research.

AI agents represent the next stage of development in which systems can perform multi-step tasks with limited human oversight. Early deployments have focused on areas such as software development assistance and workflow automation. The available studies suggest that productivity gains have not consistently translated into clear financial returns.

Industry forecasts continue to project significant economic impact from these technologies in coming years. Corporate spending on AI infrastructure and services has grown steadily. The gap between projected value and measured returns remains a subject of ongoing examination.

The studies do not conclude that no companies have achieved positive results. Instead they indicate that for the majority of deployments examined the measured ROI has fallen short of expectations set during initial adoption phases.

Key Facts

MIT study
found limited ROI for generative AI in most businesses
Additional studies
six or seven teams reached similar conclusions
AI agents
show similar pattern of hype versus measured returns

Story Timeline

2 events
  1. 2026

    Multiple studies question financial returns from AI agents.

    1 source@GaryMarcus
  2. 2023-2025

    MIT and six or seven other research teams publish studies on generative AI ROI.

    1 source@GaryMarcus

Potential Impact

  1. 01

    Further academic research into AI productivity gains is likely to be published.

  2. 02

    Companies may adjust investment levels in AI agent deployments based on measured returns.

  3. 03

    Businesses could focus implementation on narrower use cases with clearer financial benefits.

Transparency Panel

Sources cross-referenced3
Confidence score65%
Synthesized bySubstrate AI
Word count242 words
PublishedMay 8, 2026, 2:52 PM
Bias signals removed1 across 1 outlet
Signal Breakdown
Loaded 1

Related Stories

Akamai Signs $1.8 Billion Seven-Year Cloud Deal With AnthropicSubstrate placeholder — needs review
ai2 hrs agoDeveloping

Akamai Signs $1.8 Billion Seven-Year Cloud Deal With Anthropic

Akamai Technologies announced a $1.8 billion seven-year contract with Anthropic for its Cloud Infrastructure Services, the largest in the company's history. The deal was disclosed in Akamai's first-quarter 2026 earnings report. Akamai shares rose 27 percent on May 8 following the…

forbes.com
1 source
Trump Administration Considers New Oversight for Advanced AI Modelstechjuice.pk
ai10 hrs agoFraming65Framing risk65/100Rewrite inherits consensus framing of AI as an imminent uncontrolled threat, using loaded risk language and lede misdirection that foregrounds political reversal over the substantive AI capability.Click to jump to full framing analysis

Trump Administration Considers New Oversight for Advanced AI Models

Anthropic's unreleased Mythos model, capable of autonomously finding software vulnerabilities, prompted a White House shift from its previous hands-off AI policy. President Trump is considering an executive order to establish a formal review process for the most powerful systems,…

The Washington Post
Benzinga
2 sources
Nvidia CEO Jensen Huang Says He Does Not Mind Paying $8 Billion in California Taxespandaily.com
ai6 hrs agoDeveloping

Nvidia CEO Jensen Huang Says He Does Not Mind Paying $8 Billion in California Taxes

Nvidia CEO Jensen Huang stated he is comfortable with his tax payments to California while speaking at the Milken Institute Global Conference. Huang addressed the proposed billionaire tax and affirmed his decision to continue living in the state. The comments came as conference a…

FO
1 source