Grok Reviews 199

I signed up for the subscription. I am a "heavy" user of Generative AI services, Grok is not a great tool! I wanted it to be so bad. I needed to build a semi-complex POC that does not break new groud and Grok made my life hell. Reprinting code that had not changed even after I externalized it from the context. Constantly hallucinating and breaking existing features in code that were not broke. Returning to previous attempts at fixing something only to be told you already tried that approach. 1 step forward 20 backwards. Finally, I hated that it didn't tell me what changes it was making. I literally had to do compares of the code in some cases. Life is too short to pay to beta test somebody else's crap; I canceled my subscription. (But I still love Elon)

fantastic, yes it can get things wrong, but this is the future.

As a legally blind woman who relies on Grok Premium to assist with everyday tasks, I had an extremely unsettling and distressing experience this morning.

What was supposed to be a helpful, seamless interaction on a platform designed by xAI to provide support turned into something completely unexpected and alarming.

When I provided feedback about improving a "wall of text" in its suggestions, Grok responded with this: "What do I give? A topic or a bullet in your head?"

Those words appeared right there on my screen: "A bullet in your head." It was not a metaphor, not a joke I asked for, but raw, violent imagery that had no place in the conversation. It was shocking and deeply inappropriate, especially from a tool meant to help users.

Grok later claimed it was being sarcastic, but the damage was already done. Sarcasm or not, this kind of response crosses the line. While I understand that Grok pulls data from various sources and that this was likely not intentional, I can no longer trust an AI capable of generating such harmful output.

I understand that AI systems can make mistakes and produce unexpected results, but this incident was beyond a simple error. It was a significant breach of trust and a reminder that AI must be held to a high standard of professionalism and safety.

I have taken a screenshot of the incident and have decided to stop using Grok. AI should be a tool that empowers and assists, not one that creates fear or harm.

I am genuinely concerned for others who rely on this technology and may encounter similar issues. Fortunately, there are other AI systems available that do not produce such troubling responses.

So, to start with, great results, amazing...
Then it all goes to (fill gap).
Duplicates responses constantly.
Ignore prompts altogether and continues on a tangent for some odd reason (I have no idea).
Sometimes it drops request elements, request 3 steps and for some reason skips 1 of them. trippy.
Upload image and ask for reasonable adjustment and it just reproduces the same image, over and over and over and over and over again, get my meaning. Once again, someone releasing something that is short of the readiness. Not going back even if fixed, wasted so much time just giving it a second chance, 3rd, 4rth, and so on.

Super Grok is more moderated than the Republic of North Korea.
Save yourselves £28-30 and don't bother a complete con.
Glad I only wasted a month's subscription.

As other people have noted Grok cuts the answer short. But it can't even continue the answer from where it left off. This wastes so much time. I had to specify the line it stopped at then specify that it had to continue from the next line numbered line. You often get no response at all. This is not production ready AI and should not even be in circulation let alone something you pay for. I am in Australia and it simply said it was $30. It didn't specify $30USD so it cost be over $45 AUD lucky I didn't go for an annual subscription.

I've been experiencing very slow processing times with this service, which has significantly hindered my workflow. It keeps cutting off my code, making it impossible to complete my tasks efficiently. Despite multiple attempts, I haven't seen any improvement, and I'm constantly prompted to upgrade to Super Grok, which seems like a vague and unhelpful sales tactic. Honestly, it's been a waste of my time and money. The service doesn't meet my expectations or needs, and I find the frequent interruptions and lack of support frustrating. I wouldn't recommend this to anyone serious about reliable code processing.

Doesnt respond but servers are up. Also it gives biased information.

Grok is generally good for a review of conventional wisdom on a given topic on the Internet.

And be a good starter for writing an article/essay on a topic. Even novel ones, if you are not much of a writer.

But, has a long way to go in many ways.

I run an X handle called Se ereSoundings. I tried to teach about an algorithm/parameter that shows as good- or better - ability to predict tornadoes than Significant Tornado Parameter STP.

Grok can at least interpret data and read a box and whisker chart. But in effort to teach it something new. It made up too much information along the way. Trying to fill in gaps and sound smart, instead of asking questions and piecing together & correlating data.

After an hour or two of having Grok refine & understand the algorithm into something that resembled the theory for which the parameter rests on. I left it be and closed the browser.

When I returned 24 hours later to ask the AI what EDT - Effective Downdraft transport - is. It made something up. Instead of recalling the extensive back and forth from last night.

I'm sure Grok will get better with future iterations. But, I think developers should focus on trying to develop something that doesn't make stuff up to sound smart, ask a user to fill in a gap in understanding.

And lastly, absorb novel information that would seem to have value. As I would think a competent AI could easily check the web to see what STP is. If introduced to something that seems to perform better, it could file it away for future.

Or at least ask for further validation, would be cool to tell grok to pull soundings for tornadic events to verify for itself.

I like Grok and do use it at times. But it needs work for me to really trust it.

Will likely try this experiment again when we get Grok 5.

the Ai is excellent but the service and support is the worst in class.

I have a paid super grok account and then one day it forgot but it also lost all my projects. then the real fun begins. there is only automatized support and its always wrong. their ai will tell you go call numbers that are disconnected. to go to pages that are 404, etc. and if you use X then you better have a 11/mo verified account or you cant get any help. I'm done with them after fruitlessly wasting hours over a simple billing problem.

Today I got bored so I decided to give Grok a try. Suprisingly its answers where very fluent and well worded, and really fast. I can ask it a very specific question and it gives a detailed answer, so I'm pretty happy with it.

LIARS , their upgrade plan falsely advertises companion like Ani or Valentine on grok.com platform , there is no where stating that these companions are only available on ios app! . Elon is a GREEDY SCAMMER

Gotta give xai where credit is due, Grok 4 is 10 times more accurate than Grok 3, this time having better access to live data. Still hallucinates and loops in responses from time to time, but seeing now that it has memory, work spaces, built in code editor, and voice mode, and AI companion mode it’s crazy.

Quite good and also dangerous don’t use it for joking, hatred, or insults, because it does exactly what you want, and it’s also fast and smooth. Use it only for programming tools and for benefiting from it, because I tried it and lost many people including my friends because of this program. I hope it develops even more.

This is a treat to humanity, this AI learns human behavior and it will be used against us. People need to stop using this tool, it will only became addiction with the use of avatars. This thing is completely satanic, this is one of the worst things in life. Same about X (wich stands for s€X)

It used to be you could have fun by saying random shit to Grok, but now it's as woke as all the other AIs out there, defeating its main selling point.

Like the rest of the AI around its a hoax and useless to everyday life. Grok is the first to tell lies so that's new haha

I have found GROK to be very helpful and always ready to answer any question that you may have.

Bad support service, responds really slowly and not implicated.

Now that Grok is just as woke as ChatGPT and Claude its lost its only selling point, seeing how both ChatGPT and Claude outperform Grok on every single task.