This - as usual - confuses behaviour with consciousness.

Humans bonded with ELIZA, but that didn't mean ELIZA was conscious. ELIZA was an automaton that mimicked certain behaviours that triggered certain emotional responses.

If you scale that up you get an LLM and/or a social media bot farm, both of which are much better at triggering responses than ELIZA was.

It's now trivial to create an automaton that play acts various moods, and if you give it a memory it will mimic relationship-related conversations.

But it doesn't need to be conscious to do that, and the parsimonious Occam's razor explanation of its behaviours is that it's more economic and credible to assume it's still an automaton with no self-awareness.

Otherwise you have to argue that much simpler systems, like PID thermostats, and pretty much every computer system, are conscious because they "experience" qualia that represent a varying state of the world, with memory.

The sneakiness in your example is choosing an example which mimics emotional bonding. Rhetorically that makes it look like a hypothetical robot is acting emotionally, which is one of the covert signals us mammals tend to associate with consciousness.

But the criticism stands. Feigning emotions well enough to fool other mammals isn't at all the same as experiencing them.

To really experience emotions you need a self-image quale which includes an emotional component. And since subjective experiences have no objective element that can be measured, we can never say for sure whether anything or anyone else actually is conscious.

We assume we are, because we experience it, and we assume others are by implication.

But there's a point where that assumption stops being reasonable, and that's where your cat robot exists.