Support for Laio - on our way to 100% truthfulness

For the past two weeks, we have already told you a bit about how we work with?Laio?- our AI-supported editorial assistant. An important aspect that has remained unsaid so far is the reliability of the outcomes. The importance of this is obvious, not only because the creators of GPT-4 frankly admit that their model is still quite hallucinatory at times, but also because it is vital to journalism.

There are two ways we are addressing this issue now. The first is from our own ongoing development: Laio 2.0 has several additional safety valves that greatly reduce the risk of nonsense. From the process principle of human-machine-human (we see Laio a bit like the trainee who is first given an assignment and eventually will be supervised and checked), this means a further upgrade in reliability. But that's not all.

Therefore, we are also enormously pleased with the support we have received from?Metropoolregio Eindhoven. Their grant will help us enormously in this endeavor to achieve 100% truthfulness. And what's extra nice is that it also allows us to enlist the help of two AI startups:?Fruitpunch AI?and?Chunkx. One will help us improve the front end of the process, and the other will look for additional control elements on the back end.

This project will last for a year; we have our kick-off meeting next week. We'll keep you informed about the progress, but for now, the conclusion is that we will make great new steps with Laio.