Data-Driven Ideas
Turning LLMs into research-powered idea engines
At Addition we’ve been exploring what AI models like ChatGPT can do when you combine them with data. Not only can you get greater control and interpretability, with the proper setup, you can go from research to creative outputs in seconds.
A few weeks ago, we wrote a post exploring how AI systems can go from research to insights instantly.
What would it look like if you skipped that step and had the AI turn your research into ideas?
Data Driven Ideas
To demonstrate how this works, we've set up a system that uses the newly released ChatGPT API to research a topic and use its findings to generate ideas.
This AI will have two complementary functions:
It conducts research autonomously to discover insights using the Google News API. The AI will use my input to explore a range of search queries to discover articles. As it discovers helpful information, it will adjust its search queries to go deeper into the topic.
It develops new ideas based on the research findings by referencing creative examples we've curated. This data set focuses on integrated advertising campaigns, including many of last year's D&AD awards winners.
Ideas for Nike
Ok, let's see it in action.
We'll give it a client: "Nike."
We'll also give it a research topic: "The Upcoming NBA Playoffs."
First, the researches the subject to read up on the conversations around the playoffs to inform its idea.
After completing its research process, it outputs this idea:
Nike will launch a new line of basketball sneakers revolving around the theme of underdogs to coincide with the 2023 NBA playoffs. To inspire confidence and loyalty, the campaign will focus on the rising stars and underrated players who could swing a series at any moment, in the same way Immanuel Quickley of the New York Knicks has done this season. Nike will also partner with Quickley as a brand ambassador for the campaign. The line will feature bold colors and designs as a tribute to the fearless spirit of underdog players who have the potential to become champions. To engage fans and promote the campaign, Nike will launch a social media contest where fans can submit their own stories of overcoming obstacles and achieving greatness as underdogs, with the winner receiving a limited-edition pair of the new sneakers.
It then goes on to cite its sources:
2023 NBA Playoffs Schedule
NBA.com
https://www.nba.com/news/2023-nba-playoffs-schedule2023 NBA championship odds: Updated title futures after All-Star Weekend
FOX Sports
https://www.foxsports.com/stories/nba/2023-nba-championship-odds-updated-title-futures-after-all-star-weekend3 bench players who could swing an NBA Playoffs series
FanSided
https://fansided.com/2023/03/06/nba-playoffs-key-bench-players/
You catch a glimpse of how the AI autonomously researches a subject and generates ideas in this video.
Business Implications
Most brands recognize that they need to move faster to stay relevant in a world that feels increasingly hyper-accelerated by news cycles and social conversations.
By putting LLMs to work on their research, teams can get to high-quality insights, ideas, and content faster. Organizations that learn how to do this will gain the ability to harness culture, insights, and online conversations in near-realtime and use that information to power their communications in new ways that reflect their brand and creativity.
We are developing a new tool to instantly turn your research into insights, ideas, and content. If you're interested in getting early access, drop me a line by replying to this email.