I've been asked why my given AI conversation examples are up to now typically based on ChatGPT.
Well, the answer is simple.
Bard is rubbish.
And it's not just that the Bard UX is presently poorer.
To gauge its calibre, I asked it a couple of rudimentary questions about a topic in which I happen to hold deep knowledge. Indeed, it wouldn't be hyperbolic to suggest that no-one could match me for it.
Namely, this blog.
My salespodder posts.
How many blog posts had I done?
It claimed access to information right up to the present day. Telling me at that time there'd been "1,035".
Sadly, the reality at that time was in excess of 3,500.
Still, maybe this under-reporting was down to some cryptic computery way of counting. Which could be understandable. So I tried to assess Bard's know-how through a different slant.
Did it know of all my posts, particularly from way back, I wondered?
I first checked if it had my correct url.
Alarmingly, its terse response was irritatingly unhelpful.
"I'm not able to help with that, as I'm only a language model."
Then shifting angle, what it proceeded to tell me about salespodder - both the blogger (me) and the entity (my blog, as the world's longest running solution seller's blog) - was so shockingly incorrect as to practically be some kind of reverse-libel.
I shan't name names. As that would be both unfair to those inaccurately put forward, as well as being potentially disastrous info that perpetuated the glaring error.
Yet what I can safely, if chillingly, reveal, is that I was supposedly an American Marketing author (who does exist), when I am absolutely not.
Even more astounding, was the stated commencement of my blogging. Continually given as 2001. When I actually duly began Jan 2006.
And at not one, but two points, Bard even claimed "the blog is no longer active". Tears welled.
There's another salespodder.
When there fundamentally is not, nor ever has been.
Which is supposedly some kind of software house "based in Portland, Oregon". Or then again, "San Francisco, California". Or simply "the United States". Who knows.
Bard certainly didn't.
It also on occasion offers three drafts as answers.
Which was helpful when enquiring the date this fictitious code shop opened.
As it variously told me that it was founded in 2005, 2006 and 2009.
At time of blogging here, Bard's only been live a couple of months.
But even so - especially given its shambolic 'day one' Webb Telescope launch demo fiasco - the discrepancy between real and imagined is tough to swallow.
AI hallucination writ large. Right in my own little lane.
And I am not alone.
The Sales point of this shuddering experience is that currently, we can outsell others by careful use of such generative tools. On which I've blogged plenty this year already.
Yes, such chatbots are sure to get better. Also yes, AI will slide into ways to remove our daily deal drudgery. And yes, they ought bring huge bonus to aiding buyer decision-making, selling creativity and situational analysis.
But it ain't there yet.
And decidedly not Bard.