In this presentation, Ray focuses on the short-term picture for research and insights;
what will change, what will stay the same, and what are the big issues when it comes to the impact of Artificial Intelligence.
The recording of this presentation can be access via NewMR.org/Play-Again.
3. Big Picture
There will be major changes over the next couple of
years
– Reaching 10% of the insights/research ecosystem
– Disrupting perhaps 1% of the ecosystem
(i.e. over $1 billion of business)
But over the next 10 years, perhaps 50% of the
ecosystem will change
4. Agenda
AI already in use
Global Data Quality initiative
Synthetic Data
Analytics
Qual at Scale
Smart DIY
Smart Data Collection
How will we know if it works?
5. Already in Use
• Translation, transcription,
Grammarly, satnav, Siri etc
• Adaptive surveys
• Fraud detection & routing
• Algorithmic approaches
• Sentiment Analysis, e.g. social media research
6.
7. Short term
• Small improvements in fraud protection, quality checks, questionnaire checking, and more
engagement
• No drop in the use of panels – no apparent change to the ecosystem
Longer term
• If AI can’t deal with the quality issues, we will see large drop in the use of panels
• If AI can deal with the quality issues, panels will be stable, but they will decline as a share of all
research
• There is a limit to the number of people who will join panels and a limit to how many
surveys they will participate in
• Synthetic data will increase
8. Synthetic Data
• Missing data – e.g. Hierarchical Bayes
• Predictive eye-tracking
• Created Personas
– Qual
– Quant
• Direct modelling
9. Synthetic Data
Short term
– Early adopters won't wait for deep validation
– Lots of examples of adoption, but a small percentage, perhaps 1%
Longer term
– It will be faster and cheaper, and if it is deemed good enough, it will
be massive
– It took the internet over 10 years to become dominant (in the
developed markets) – this could be true for synthetic data
10. Analytics
Qualitative analysis
– Same data, but faster
– More data (qual at scale)
Quantitative analysis
– Developing slower than Qual
– Fixated on explicability and repeatability
11. Analytics
Short term
– Early adopter clients - 1% disruption
– Agencies using it to speed up or deepen qual analysis,
no visible disruption
Longer term
– More open-ended questions asked
– More qual data used in analysis,and more formal analysis approaches
used (e.g. Grounded Theory, Narrative Analysis,Discourse Analysis)
– Human/AI tools for analysis of quant and qual
– None of these changes leading to a big disruption, generally quality
improves, speed improves, prices fall
12. Qual at Scale
A special case of Analytics
Includes:
– Analysis of social media
– Chatbots to collect the information
– Quantitative semiotics, ethnography, and cultural analysis
Bifurcated epistemological position
– Operationalising qual into quant
– Large qual
13. Qual at Scale
Short term
– Projects might move from one supplier
to another, but change is early adopter driven, no net
change
Longer term
– A shift from traditional quant to more non-numeric
inputs
14. Smart DIY
All aspects of DIY will be AI embedded
– Design, fieldwork, analysis, & summarising
Democratisation of research
– More people conducting research
– More safely
– More research being conducted
15. Smart DIY
Short term
– Explosion of tweaks to existing products
and lots of new tools
– Most new products will struggle to break through
Long term
– Nearly all DIY/Self-serve will be AI aided
– DIY/Selve-serve will grow and grow
16. Smart Data Collection
Continued development of
– Adaptive scripting
– Interactive bots
No big changes
– Just a gradual shift to Smart Data Collection
17. How will we know if it works?
• It will mostly be quite hard
• LLMs are very plausible
• Establish counterfactuals
– Such as test and control
18. 3 Thoughts
1. Even fast change takes time – perhaps 10%
disruption in 2 years
– And 50% disruption in 10 years
2. AI is mostly going to be about faster and cheaper
– Only occasionally better, often worse
3. Consider Ethics
– Not everything that is possible, is legal, not everything
that is legal is ethical