OPINION27 November 2023

‘AI will disrupt every phase of research’: Views from the ASC conference

AI Opinion Trends

Generative AI was the focus of last week’s Association for Survey Computing (ASC) conference, with speakers exploring how it is impacting the insights sector. JT Turner offers an outline of the day’s perspectives.

yellow lightbulb floats above blue ones against a blue background

Last week, the ASC held its second in-person and live-streamed event of 2023, ‘Generation Survey’. This full day event, chaired by AJ Johnson of Mediaprobe, looked at the impact of large language models (LLMs) and generative AI within the survey industry.

The event included a wide range of papers by speakers from agencies, brands and technology firms, showcasing their experiments, applications and work using generative AI.

The event’s keynote was from Steve Phillips, founder and chief executive of Zappi, who shared his vision of the future, discussing his thoughts on how generative AI is impacting the insights industry and predictions on how the research process could change.

He commented: “Clients are both incredibly excited and incredibly scared. Change [from AI] is going to happen much quicker than the internet, and much quicker than mobile.”

Phillips continued that “AI will disrupt every phase of the research process” and said LLMs and synthetic data could “massively reduce the number of questions we ask, moving to more open questions and chat interactions”.

AI will also impact the structure of client-side insight teams, as well as and their data pipelines and data infrastructure, he added.

Jérome Sopoçko of Ipsos Askia gave an introduction to generative AI entitled ‘Math not magic, a dummy’s introduction to large language models’. He predicted the future of research jobs: “You will not be replaced by AI but you will be replaced by someone who uses AI and can write prompts really well. It’s a skill you need to acquire now.”

Jack Wilson from 2CV shared the agency's learnings and challenges with AI-assisted research, demonstrating how it used AI to assist with survey design, data collection, analysis and reporting.

Alexandra Kuzmina of MMR Research Worldwide explored the potential of using AI-generated avatars to interact with survey respondents, evaluating if they can improve the survey experience and data quality.

Mohamad Hussien from the BBC presented some of the methods and tools that the broadcaster uses to analyse open-ended global survey responses. He shared real examples and code demonstrating topic modelling and text summarisation to understand audience feedback.

Andrew Le Breuilly of technology consultancy Arrowstream argued that to leverage the power of AI fully, survey data needs to be rich in context, with meta data such as time, location, behaviour and emotion. He suggested ways of capturing and integrating contextual data in surveys and the use of standards such as TS-API.

David Wright of Hello Ara presented practical ways to use generative AI to improve the research process, such as generating hypotheses, testing assumptions, exploring alternatives, synthesising information, and communicating insights.

Signoi’s Andrew Jeavons explained how large language models can be used to predict how respondents would answer survey questions, and how this can help with survey design, quality control and analysis.

Industry veteran David VL Smith of DVL Smith and Adam Riley of Decision Architecture presented ‘Extinction is the rule, survival is the exception’. They discussed the new skills and competencies that researchers need to develop to survive and thrive in the age of AI.

The event concluded with a panel discussion around the ethics of AI with Debrah Harding from MRS, Lexy Kassan from Databricks and Phil Sutcliffe from Nexxt Intelligence. The group discussed a wide range of topics from the potential for participants to be exploited to the impact of AI legislation and synthetic data.

Debrah Harding commented: “For all the technological developments [we’ve seen], there is still a need and a desire from clients to see and hear real people.” She continued: “Synthetic data has a lot of potential but is still based on participant-led data.”

The event coincided with the launch of the new MRS guidelines on the use of AI, which provide a framework for ethical and responsible use of AI in research. The guidelines cover topics such as transparency, accountability, privacy and security, and offer practical advice and examples for researchers.

JT Turner is chief executive of Delineate and ASC director

0 Comments