Posted by: structureofnews | February 5, 2023

Right Tool, Wrong Job

By itself, such a system based on algorithmic rules has some real limitations, as we learned – and not just with the language generation part of it. Gary Marcus again:

The weakness of the symbol manipulation approach is people have never really solved the learning problem within it. So most symbol manipulation stuff has been hard-wired. People build in the rules in advance. That’s not a logical necessity. Children are able to learn new rules. So one good example is kids eventually learn to count. They’ve learned the first few numbers, 1 and 2 and 3. It’s kind of painful. And then eventually they have an insight, hey, this is something I can do with all numbers. I can just keep going.

And the kind of brute force data analysis we were pushing Lynx Insight into was always going to hit a ceiling at some point. But pairing even something like that with an output generator like ChatGPT? That would have been a very interesting exercise, not least if we wanted to produce, at scale, more customized versions of similar stories, based on verified data analysis. Factual journalism, in other words.

And to be sure, there are smarter AI systems out there than what we used with Lynx Insight; Anthropic, which just announced a deal with Google, seems to be streets ahead of ChatGPT in terms of its ability to “understand” broader language concepts and have more of a model of the world. (At least, based on one very quick demo that I saw, and I was drinking at the time.)

All of which is mostly an argument – or plea – here that we think more holistically about how AI can help or reshape journalism, beyond a single capability, miraculous though it may seem.  ChatGPT does a couple of things well and other things badly. Other systems can fill in the gaps.  It’s less important to criticize it for shortcomings it wasn’t designed to address and more important to think about more holistic systems that can help us do our jobs better, serve communities better, and make our products better.

In other words, to stop just looking under the ChatGPT light – bright though it is.


Pages: 1 2


  1. […] for Semafor (you know, where I work) about the possible uses of AI chatbots in journalism.I wrote earlier here about how we keep trying to shoehorn their capabilities into things the don’t do well; this […]

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s


%d bloggers like this: