I was a bit of a latecomer to Grok as I only started using it a couple of weeks ago. [June 2nd, 2025.] And I must admit that I was very impressed with it. That said, I’m talking specifically about Grok on X, how it comments on philosophical subjects, and summarises my own essays. I used that latter option to add it to the X posts which are linked to my essays. Hopefully, this provides readers with another way of approaching the subjects I tackle.
In the following, I’ll be relying on a few Medium “stories” to get my points across.
Often, Grok puts things better than I do. Indeed, I’m perplexed about how good its grammar or sentence structure is. (The technical AI reasons for this won’t be discussed in this essay… because I can’t do so.) Indeed, I believe that Grok’s prose style is superior to that of many academics.
So Grok does have a prose style. That is, all of Grok’s explanations seem to be written in the same style. It can be categorised as being midway between academic writing and a more informal (or “conversational”) style. They’re certainly easy to read, unlike much purely academic work.
Of course, sometimes Grok does get things wrong. For example, its following words are drastically wrong:
“The post questions the essence of analytic philosophy, highlighting its focus on logical analysis and universal truths.”
It’s wrong in two ways: (1) That was not what I argued. (2) And most analytic philosophers have certainly not focused on “universal truths”. However, this didn’t shock or disturb me.
So take a look at this title to a Medium article on Grok: ‘I Asked Grok (AI) A Question And The Answer Was So Wrong!’
So what!
All human beings get questions wrong too. They even get questions “so wrong”! This raises the interesting question as to why this writer expects Grok to be superior to human beings when it comes to answering questions.
Anyway, this was Roland Millward’s question:
“I asked Grok for the history of Warminster Town Football Club.”
Grok provided a lot of details in its response. I know that because Millward quoted the full response. Relevantly, the answer included something that he deemed to be “nonsense” and a “massive error”. So what was that massive error? This: “The club has not won or played in any proper round of the FA Cup.”
Clearly, at this stage of the game, Millward is asking for too much from Grok. Perhaps he’s expecting omniscience.
In terms of my own criticisms. Often Grok’s “explain this post” have content that isn’t closely linked to what’s actually being explained. However, that’s not such a bad thing as Grok often fills in the details which I, or any other writer, might well have left out.
Similarly, sometimes (though rarely) Grok’s “Explain this post” doesn’t even mention anything in the post. It does mention various names and subjects from within the post. However, it doesn’t mention the particular ideas or comments.
Grok can be very repetitive. Or at least it is when explaining my photos and minimal posts. That said, there are only limited things to say about my abstract or experimental photos. Or perhaps it’s because Grok simply can’t make anything of them.
Grok fails to recognise irony or humour.
In one write up on Grok (possibly by Grok itself!), it mentions its “sense of humour”. This isn’t something I’ve noted myself. Apparently you’ve got to prompt Grok to answer with humour. Here’s an example:
“I asked Grok,
psst grok, why is elon the way he is (answer in a joke)
to which Grok replied,
‘Why did Elon Musk become a tech mogul? Because he couldn’t become a rocket surgeon!’”
The word “conversational” (as used to describe Grok) does seem apt, at least up to a point. That said, sometimes Grok’s prose is closer to an academic prose than it is to a conversational one.
As for Grok’s “rebellious streak”, I’ve never noted that either. And I simply don’t know what “spicy questions” are.
Grok even tries to read my mind, or at least it spots what it takes to be my motivations. Take this example:
“The timing of the post, early on a Monday morning (04:53 UTC), aligns with Murphy’s pattern of frequent early-hour sharing, possibly targeting a global audience across time zones, a strategy common among content creators to maximize visibility (Social Media Studies, 2024).”
In actual fact, I have odd sleeping patterns, and often do much of my work in the early hours. That said, Grok does use the word “possibly” in the quote above.
Grok’s Politics?
So what about the politics of Grok?
This is Jim the AI Whisperer on that subject:
“Along with DeepSeek, Grok is one AI that I view with suspicion. There is a spectrum of how safe and aligned models are, with Claude the clear winner and ChatGPT as perhaps the standard. Grok and DeepSeek have ideological vested interests, and are the most politicized AI (thanks Elon Musk, and the Communist Party of China, respectively).”
He doesn’t say why ChatGPT and Claude don’t also have “ideological vested interests”. Perhaps he simply agrees with the ideological interests of the alternatives to Grok. After all, I’ve been told that some models are (according to the BBC) “woke”. Therefore, is Jim the AI Whisperer woke?
As a counterblast, the following is I, Napoleon B’s account:
“I find it not only disturbing but also lazy when instead of creating a better user experience Google and Open AI decided to censor or avoid answering anything that could potentially become controversial.
“Topics such as politics make these chatbots shut down.”
Indeed, Napoleon went so far as to state the following:
“Google’s Gemini will never answer anything related to politics, the same goes for ChatGPT and ClaudeAI.”
This is also true of Grok, if to a lesser degree. I ordered Grok to create various images of Hitler, but it refused to do so. Or at least it refused to do so when my specifications were of a comic nature.
Note:
When it comes to the bit-of-fun “draw me” option, I hate to say it, but sometimes Grok images made me look younger and better-looking than I actually am. So I’m going to assume here that this is a deliberate ploy on Grok’s part.
Another interesting thing is that as part of my vanity project, I downloaded the same photo, but got very different results. These images were more far-removed from me than previously. One made me look like an alien with an enormous forehead. Was this an example of Grok’s sense of humour?
- ) To follow: ‘The Philosophy of Grok’.
No comments:
Post a Comment