As MD of a company that has been in the online customer service service field for over a decade and has a Virtual Agent (chatbot, if you like) product, I agree with most of what Andrew Orlowski writes.
It has never been helpful to associate chatbots with Artificial Intelligent. They just aren't. We never do do so and it's a key point of standard customer presentation deck to differentiate the two things. Interest in chatbots has grown due to the rise of Siri, Cortana, Alexa et al and it's lazy shorthand by journalists just to slap an "AI" label on these, just because they make a token stab at communicating conversationally.
In fact, if we want to blame someone for instigating all of this, I suggest... Alan Turing. The famous Turing Test effectively invites people to draw an equivalence between AI and conversational interfaces. It's a false equivalence. The goal of machine learning should not be to mimic human beings. There are plenty of other computing problems to be solved without such red herrings. I suppose he did OK at some other things, despite that epic fail.
We're concerned that the escalating hype around AI will lead to raised expectations around chatbot interfaces that can't possibly be met.
That said, chatbots do have some benefits over more conventional interfaces. In particular, they increase engagement with a wider demographic of user by making it easy for the user to communicate their request in their own way. Everyone, from a very early age, knows how to ask a question and get an answer. We have plenty of data that suggest that users, overall, enjoy the experience.
Try thinking of chatbots as an alternative interface that everyone can use but that not everyone needs to.