1 is because it will spook investors, not the public.
Historically people are very reluctant to directly pay for them and in any case providers might want to double dip on ads.
Your local used car dealership knows a lot about cars. It's their business to do so. Still, they are highly motivated to not tell you the whole truth and many people know that (some don't)
My prediction is that ChatGPT will be exactly like that or in other words: Smooth talking Google which is untrustworthy beyond "factual" searches like "biggest X", "weight of Y", "president of Z"
There's little value in simply showing an ad, all the value is in the conversion. Yes, Google is well positioned to destroy that conversion rate, but that comes at great harm to itself. Display ads go for 1/10 the price of a search ad, and I'd wager text would be worth even less than images, and I'd bet the tolerable amount of ads per query is significantly higher for search as well.
Is Google willing to bet the house that (2) is a better business model than the clicking described in (3)?
Or models that run in the browser to detect and remove product placements from the text itself. Though that seems very unlikely.
1. Google already has their own internal equivalent of ChatGPT [1]. They haven't deployed it publicly yet, probably out of fear of bad PR.
2. Unlike search ads, which you can block or ignore, chat ads will be woven into the conversation at opportune times.
3. A chat format means users will spend more time on Google, rather than clicking on a link and leaving -- which means more opportunities for ads.
[1] https://blog.google/technology/ai/lamda/