The controversial role of AI in foresight work

In 2019, if you’d ask ten people working on foresight and future trends how many of them use AI in their work daily or weekly, the number wouldn’t have been very high.

Today it’s a different story. The bleeding edge technologies have become the tools we use today, but obviously not without controversy. The market is getting saturated with both AI-powered tools and AI-augmented tools. Sometimes it feels like there are no "dumb" tools left.

How we approach artificial intelligence technologies

At FIBRES, we’ve been actively scouting the emerging AI technologies, experimenting with them, and incorporating the best of the breed into our product’s features. We have been mindful with AI, only utilizing it where we see the most benefit and the best fit for purpose.

This has led to us integrating technologies such as large language models (LLM) and diffusion models as well as more "traditional" artificial intelligence such as natural language processing (NLP). However, this has not been without debate and deep reflection.

Can we trust AI? What can we trust it to do and what is left for humans? Where does the data come from? Are we bursting bubbles or just creating more of them?

Those are just some of the questions we keep asking ourselves—and that our customers also ask us. But we think we found a balance early on and have kept it since, and are pretty proud of it as well.

Human-centric philosophy on utilizing artificial intelligence

Our philosophy on utilizing AI technologies boils down to these five statements:

1. Use AI to tackle the painful, boring, repetitive, or time consuming tasks, not to take the joy out of foresight work from humans.
2. Enable humans to achieve more and greater things than they would be capable of without the AI-powered tools.
3. Be purposeful in utilizing AI, seamlessly incorporating it in purpose-built features and functionalities. No AI for AI’s sake.
4. Never mislead humans to think AI always has the correct answers or is telling absolute truths.
5. Always show what is AI-generated and what is human-made or human-verified.

With our recent upgrades to our AI capabilities, we’re facing many of the same questions again, but also giving better than ever answers to the big questions.

As an example, our newest custom trend dataset builder is able to pull in sources from the web, draft trend descriptions based on the sources, and plot everything on a fully customizable trend radar. It's taking heavy workload off the human.

By making the sources visible and linking to them from the trend descriptions, we’re making it super easy to verify the sources and the AI-generated trend descriptions. Human verification is essential and made easy.

The custom trend dataset builder produces so convincing results that without warning you might think someone has done days and days worth of work collecting signals, writing trend analysis, and visualizing it all on a radar. That’s why we’re guiding all of the users of this service to take it as a starting point—a draft of a full trend radar—and then verify and build on it to make it their own. We explicitly mark the drafts as generated by AI, so humans are not mislead into thinking the content has been human-curated.

Powerful tools require control

We see the role of AI in strategic foresight is rapidly evolving, and we’re doing our best to leverage these advancements for the benefit of the foresight practitioner and their organization. We work to take these powerful technologies and incorporate them into FIBRES in such a manner that the risk to benefit ratio is optimal.

You can definitely go wrong with AI if you don't understand it or if you misuse it. It's well documented that certain AI technologies may have biases, they may produce hallucinations, and they can even have tendencies to gaslight the user. That's why rigorous testing and control is essential and has to be done continuously after the initial identification and validation of the various impressive AI use cases.

How do you see it? Has AI changed the way you work on foresight for the better? Are there further disruptions or opportunities ahead?