UK Politicians Test AI Focus Groups as Firms Face AI Cost and ROI Scrutiny
UK politicians test AI-generated focus groups as firms struggle to control AI costs and prove returns.
Startups Electric Twin and Artificial Societies, along with established research firms GWI and Toluna, are using data from real focus groups and surveys to build AI personas that can respond to political messages, according to Bloomberg. London-based Synthetic Respondents charges $16 (£12) per project, which it says is one-thousandth of the cost of a traditional piece of focus group research.
The political experiments are taking place as companies face harder questions about AI spending. Forrester's State of AI Survey 2026, cited in a TechRadar report, predicts AI will account for one-fifth of all business technology spending in 2027. Yet more than half of the CIOs surveyed expect to miss their AI ROI targets next year.
About 45% of CIOs want to see ROI within one year to consider their rollouts a success, the survey found. Over the past year, just 22% of AI decision makers reported a positive impact on earnings, according to the TechRadar account of the Forrester research.
Most tech leaders expect to increase budgets next year, but only 53% expect the increase to exceed inflation. Forrester also pointed to rising AI consumption costs, such as per-token charges, as providers shift from seat-based subscriptions to consumption-based models. It expects cheaper open-weight models such as Gemma, Llama and Qwen to become more popular for enterprise workloads that generate the most token volume, including document processing, customer support conversations, internal knowledge search and routine substeps inside agentic workflows. Forrester said a meaningful quantum computing breakthrough could come as soon as next year and add further pressure to budgets; 87% of businesses familiar with quantum computing have no resources allocated to it.
A separate Splunk State of Observability 2026 report, also cited by TechRadar, found that 55% of organizations cannot effectively control their AI costs, and 57% worry that telemetry costs are higher than the infrastructure they are monitoring. Three in five said unexpected spikes in AI bills cause more financial strain than a consistently high baseline, suggesting a preference for flatter pricing.
Splunk found some operational benefits from AI monitoring. Four in five organizations, or 81%, said monitoring AI workloads increased collaboration between observability and security teams, and 73% can accurately trace problems to their underlying security causes, up from 65% a year earlier. Among observability leaders, 91% said they can attribute AI costs to a particular model, agent or workload, compared with 69% of non-leaders.
Forrester expects businesses to become better at linking AI investments to actual financial outcomes, while Splunk's data shows that cost attribution is stronger among observability leaders than among other organizations.