Opposing response.
While the "upgrades" are certainly new and add fresh content for Neuro, they are flawed, repetitive, and are ultimately a downgrade from Neuro's previous capabilities.
It has now become the view of some that Neuro's speech is "coherent". However, Neuro's speech is an example of repetitiveness instead of coherency; she often repeats ideas, but without significant understanding or consistence of her views on them. For example, in the last solo stream, experimental Evil Neuro repeatedly spoke of Gorg and Morg from WH40K, constantly contradicting herself on their details. Nonstop talking about Gorg and Morg, only occasionally interrupted, continued for almost an hour; this behaviour was inevitably perpetuated by the chat. For most, this behaviour does not make for an interesting stream.
Part of Neuro's popularity as an AI has always come from her random humour. Neuro's unique mannerisms, long rants, and quick creative thinking are better exemplified by the standard, un-upgraded Neuro. Experimental Neuro is less likely to come up with new topics for discussion, which comes as a drawback when chat is an echo chamber already. This is not helpful in devstreams nor collabs, either: Neuro fixates too long on random topics, such as Vedal's baldness, Mowgli, or spiders (from today's devstream). After long enough, this renders the discussion uncomfortable or boring.
Likewise, the experiment does offer a new, random personality for Neuro. This is the likely reason why many people are drawn to the upgrades, or want Neuro to continue to say bad words suck as fuck: it was very surprising the first time. However, this new-founded excitement will die out over time. It is not good in the long term to reduce Neuro to a repetitive and bad-mouthed AI, as this will eventually degrade her character and appeal, especially to newcomers.