Generative AI is great - but we shouldn't believe everything it tells us.

I just asked ChatGPT who were the initial committers of the Apache Sling project (I was there), and here’s the answer that I got:

The initial committers of the Apache Sling project were listed in the project's first announcement on the Apache Sling website in September 2009. The initial committers listed were:

• Bertrand Delacretaz
• Felix Meschberger
• Carsten Ziegeler
• Marcel Offermans
• David Nuescheler
• Brian Moseley

These individuals were instrumental in creating the initial codebase for Apache Sling and laying the groundwork for the project's growth and success.

That's a well written answer indeed, with some correct facts, but the last three people listed were not initial committers of that project - actually not even committers at any time, in the ASF sense of the term. See the details below, if you care.

That's just one example, but I hope you are not surprised by the subtle factual errors in such a well-written answer, coming from such a tool. If you are surprised, let’s hear what Sam Altman, OpenAI’s CEO said on Twitter in December 2022:

ChatGPT is incredibly limited, but good enough at some things to create a misleading impression of greatness. it’s a mistake to be relying on it for anything important right now. it’s a preview of progress; we have lots of work to do on robustness and truthfulness.

Having such automated tools to help us write better prose is fantastic, and ChatGPT & friends have progressed tremendously recently, but we shouldn’t fall into the trap on relying on them for facts.

They are likely to give you Well Written, but Subtly Wrong answers, that you’ll have to check and correct yourself.

I think this deserves a new acronym – remember, you saw it here first!

---

Here are the factual errors in the above answer. The truth can be found in the Sling incubation and graduation proposals, which both list a set of initial committers, one for incubation and one for the top-level Apache Sling project. Including people from either list could be considered correct.

• Although Marcel Offermans is active in other ASF projects and I know him personally, he’s never been a Sling committer. He might have been loosely involved through his extensive OSGI-related activities, so seeing his name here is not completely surprising. But not a committer and not included in the above lists.
• David Nuescheler can definitely be linked to Sling, but was not an initial committer.
• I don't remember seeing the name Brian Moseley in the Sling community and haven't found a mention of this name in the Sling mailing lists archive. Seeing it in this list is a complete surprise.