Via @rodhilton@mastodon.social
Right now if you search for “country in Africa that starts with the letter K”:
-
DuckDuckGo will link to an alphabetical list of countries in Africa which includes Kenya.
-
Google, as the first hit, links to a ChatGPT transcript where it claims that there are none, and summarizes to say the same.
This is because ChatGPT at some point ingested this popular joke:
“There are no countries in Africa that start with K.” “What about Kenya?” “Kenya suck deez nuts?”
AI generated content is pure search result pollution.
What Google should be doing rather than pushing Bard is detecting AI nonsense and purging it from their search database.
You cannot detect AI generated content. Not with any real world accuracy and it will only get worse.
Also, because google relies on growth for everything from compensation structure to business model, they are in a bind - ads is not growing anymore, it’s done.
And while they managed to create an illusion of growth this earnings round by juicing subscription fees 20% and increasing ad load everywhere, it’s not a sustainable tactic. We are already seeing a tech sell off as people are getting less and less secure.
So they rely on AI narrative to keep investors invested Google needs AI to work or the investors will move it to a place that may offer higher returns than a squeezed out ads model.
Worse even they are being attacked by AI - on the quality front (junk content) and in the marketplace (openAI), they don’t have a choice but to take a pro AI stance.
AI is its own worst enemy. If you can’t identify AI output, that means AIs are going to train on AI generated content, which really hurts the model.
Its literally in everyone’s best interest, including AI itself, to start leaving identification of some kind inherent to all output.
Those studies are flawed. by definition when you can no longer tell the difference the difference on training is nil.
It’s more like successive generations of inbreeding. Unless you have perfect AI content, perfect meaning exactly mirroring the diversity of human content, the drivel will amplify over time.
Given chinchilla law, nobody in their right mind trains models via shotgun ingesting all data anymore. Gains are made with quality of data at this point, less than volume.