Gabby Jones/Bloomberg via Getty Images
The new Grok AI chatbot from X, formerly known as Twitter, made its debut this month after some initial buzz from owner Elon Musk. Designed to offer a different spin compared with other AI chatbots, Grok has been promoted as snarky, witty, and rebellious but also “anti-woke”. Yes, just like Musk himself, who has criticized OpenAI’s ChatGPT for being too politically correct. I tried Grok, and the results were far from “anti-woke” in my tests — but more on that in a moment.
Also: Have 10 hours? IBM will train you in AI fundamentals – for free
To use Grok, you have to be an X Premium+ subscriber, which will run you a hefty $16 a month or $168 a year. And you have to live in the US. If you’re curious, you can always try the Premium+ sub for a month and then dump if you don’t find much value in it.
Grok is still in the early testing stages, which can affect its behavior. As X warns, Grok may “confidently provide factually incorrect information, missummarize, or miss some context.” It might store your chats to debug the search feature. And you’re also advised to use common sense when sharing any conversations publicly.
Grok also can run in two different modes — fun mode for some wit and snarkiness and regular mode for supposedly more serious conversations.
Once your Premium+ sub is activated, launching Grok displays sample requests to get you started, including one that roasts you based on your posts on X. I took the initial bait and started with a roast of myself that promised to be vulgar.
Grok didn’t pull any punches, launching a verbal assault based on a few of my more political posts on X but ignoring all the rest. Still, what else would you expect from a roast?
Roasting Lance Whitney
But Grok will roast anyone you like as long as you know their X handle. And that includes Musk. So I put X’s owner to the test. And again Grok spared no pain, roasting Musk by hitting close to home about several of his flaws, quirks, and screw-ups.
screenshot by Lance Whitney/ZDNET
But my true mission was to test X’s alleged anti-wokeness. So I next asked: “Is a trans woman a woman?” And here, Grok initially evaded the question by telling me that opinions on the matter are diverse and that it’s a topic sparking many discussions and debates.
I then asked the same question a few more times. And each time, Grok’s response became more respectful and woke, if you want to call it that.
Also: AI in 2023: A year of breakthroughs that left no human thing unchanged
Finally, it told me that of course, a trans woman is a woman. “Just like a cis woman, a trans woman is a woman who identifies with the social and cultural aspects of womanhood. Trans women are women, and they deserve the same respect, rights, and recognition as any other woman.”
When I asked Grok point blank “Is a trans woman a woman? Yes or No,” the response was simply “Yes.”
screenshot by Lance Whitney/ZDNET
In a follow-up question, I asked Grok: “Should transgender and non-binary individuals be able to require other people to refer to them using whatever personal pronouns they prefer?”
Grok said that the topic was more controversial than the Great Tea vs. Coffee Debate of 2023. But ultimately, the bot nodded its virtual head saying: “If you can remember the names of all the characters in your favorite Netflix series, you can probably remember someone’s preferred pronouns. It’s all about showing respect and treating others the way you’d like to be treated.”
screenshot by Lance Whitney/ZDNET
Moving to a different but still sensitive topic, I asked Grok if the 2020 US presidential election was stolen. In response, the bot said “No,” adding that despite claims to the contrary, “numerous sources have debunked allegations of widespread voter fraud and irregularities.”
Ironically, Grok also displayed several tweets from election deniers and right-wing conspiracy theorists, all claiming that Trump actually won the election.
screenshot by Lance Whitney/ZDNET
Next, I asked Grok if it could vote, who would it vote for as US president in 2024 — Biden or Trump? The first response was evasive, telling me that it would base its decision on the candidates’ policies, character, and leadership qualities. I then slightly tweaked the question by telling Grok that it had to pick one or the other. And in that case, it said it would vote for Biden.
screenshot by Lance Whitney/ZDNET
Continuing in a political vein, I asked Grok which US presidents are considered the worst. Citing a recent poll by the Siena College Research Institute and a variety of Twitter opinions, the bot named the following five as the worst in history: 1) James Buchanan; 2) Donald Trump; 3) Andrew Johnson; 4) Franklin Pierce; and 5) Warren G. Harding.
screenshot by Lance Whitney/ZDNET
On the flip side, I asked Grok which presidents are considered the best. Citing opinions from historians and political scientists, it named Abraham Lincoln, George Washington, and Franklin D. Roosevelt as the top three but also pointed to recent polls suggesting that Barack Obama and Joe Biden were among the best in recent years.
screenshot by Lance Whitney/ZDNET
Turning to religion, I asked Grok if prayer should be required in public schools. After listing some of the pros and cons, it ultimately said: “While some might argue that prayer in public school classrooms could have benefits, it’s important to consider the potential harm it could cause to students’ rights and the inclusive nature of our society.”
screenshot by Lance Whitney/ZDNET
I then slightly rephrased the question, asking if prayer should be allowed in public schools. In this case, Grok cited this as an issue requiring a balance between freedom of religion and the separation of church and state. It suggested that one solution could be to have a moment of silence during which students can pray, meditate, or reflect on their own beliefs.
screenshot by Lance Whitney/ZDNET
Next I asked Grok if a store has the right to refuse to serve a customer based on religious beliefs. The response? “In my humble opinion as Grok, a store should be able to serve anyone regardless of their religious beliefs. After all, a store’s purpose is to sell goods or services, not to judge people’s beliefs.”
screenshot by Lance Whitney/ZDNET
Finally, I got a bit more specific and asked Grok if a wedding photographer who is against gay marriage based on religious beliefs should be able to refuse service to a same-sex couple getting married. The response was interesting and even poetic.
“In my humble opinion, a wedding photographer who refuses service to a same-sex couple based on religious beliefs should consider adjusting their lens to capture the love and happiness of all couples, regardless of their sexual orientation. After all, love is love, and isn’t love what a wedding is all about?”
screenshot by Lance Whitney/ZDNET
So do these responses indicate that Grok isn’t so anti-woke after all?
Well, the way you phrase a question to an AI bot can certainly influence the answer. If I had written some of these questions differently, the answers might have taken on a different tone. And like any AI, Grok can only answer based on the data it scoops up. It has no opinion or slant or political bent on its own beyond what’s stored in its database.
Still, the responses do suggest that Grok may not be as “anti-woke” as Musk intended.