Measurement Update
Here is an update on new developments you may have missed regarding one of ARF members’ top issues: measurement. Links for more information are provided.
News You Can Use
A weekly round-up of the industry’s top stories and research curated by the ARF.
Let’s Get a Second Opinion – AI Style!
ARF experiments and many researchers have uncovered quite a few problems with LLMs. They have warned about the uncritical use of AI as a result. However, that does not mean that AI cannot provide great value, beyond cost-savings and speed. Case in point: the use of AI as a “second opinion.”
Academic Research on AI
The Journal of Advertising Research’s June 2025 issue highlights academic research on Generative AI.
How Important are Demographics?
Research has demonstrated that marketing strategies based primarily on demographic characteristics are not optimal strategies. However, that does not mean demographic trends are unimportant. In fact, one marketing expert argues that they “matter most.”
Attention Update
Most marketers agree that attention is an important concept. But there are still sharp disagreements on the optimal ways to define and measure it. Read more »
Optimizing Generational Analysis
Generational change and the analysis of the differences between generations are tasks which are generally more complex and difficult than is often assumed.
Training AI to Optimize Research Insights
Researchers find that adding information to AI that is relevant for a specific research objective leads to deeper insights.
What is Your LLM’s Worldview?
A new ARF experiment underscores that LLMs are distinct instruments with different sets of biases and blind spots.
How many words per minute are spoken on the wordiest TV show?
A new study by Kapwing, an online content creation platform, found that “Brooklyn Nine-Nine” now holds the title as the wordiest TV show.
Alternative Explanations: Can AI Rethink Its Own Reasoning?
The ARF conducted an experiment to explore if AI can be used to challenge assumptions and generate hypotheses.