Is Grok Worth Using?

Grok's new voice update promises faster, smoother, smarter conversations. So I tested it on English practice, pronunciation, live search, and a story game and paid $9.90 to find out how it holds up. ✅ Check out the 90-day program: https://www.lukepriddy.com ____ 0:00 What Grok's new voice update promises 0:56 Conversation practice: does AI hurt creativity? 1:29 Asking Grok what people on X actually think 2:07 My answer, and Grok's first correction 3:03 Asking for two or three more corrections 3:30 Starting a pronunciation test, and hitting the limit 4:31 Paying $9.90 to keep going 5:11 Teaching me the word "usually" 6:01 "Are you sure that's right?" 7:07 Four syllables, three syllables, and a loop 7:43 Is this really speech to speech? 8:26 The honest answer: it transcribes first 9:15 Live search: tech events in NYC this weekend 9:38 The AI hackathon breakdown 10:25 Finding a falafel spot near the venue 10:55 A back-and-forth story game 12:14 Grok Bot: a different take on AI tools 13:25 Bots that manage other bots 14:41 One agent who knows you, running the subbots 14:59 The new app, and what I still need to test 15:12 My honest verdict on Grok voice 15:31 Wrap-up and the 90-day program ____ Grok just shipped a voice update that's supposed to be much better: smoother, smarter, more natural conversation. So I ran it through the things I actually care about — conversational English practice, pronunciation teaching, live search during a conversation, and a bit of creative play. We start with a real discussion about whether AI hurts creativity, and I ask Grok to correct my grammar as we go. That part works. It catches "before I answering this question" and "which are very interesting," and explains why. Then I hit the free voice limit almost immediately, pay $9.90, and keep going. The pronunciation test is where it falls apart. I ask it to teach me "usually," and it says "usual Ali," then insists it's saying three syllables while clearly saying four. It apologizes, tries again, and gets stuck in a loop. I also ask it to confirm whether the new Think Fast 2.0 mode is true speech to speech — it says yes, then corrects itself: there's still a transcription step in between. That explains a lot. Live search is stronger. It finds an AI hackathon at the New York Public Library with times, location, and what it's about, then finds a falafel place a ten-minute walk away with a price range. The story game is fun too, and ends better than I expected. At the end I look at Grok Bot, which is a genuinely different idea: you build bots that act on their own and coordinate each other, instead of chatting with one assistant. One learning coach managing specific subbots, connected to Drive and your calendar. I'm still testing it, and if it's interesting I'll make a separate video — tell me in the comments what you want me to try. The voice verdict: it sounds natural, but it just doesn't seem that smart. I don't think it's worth it yet. #learnenglish #english #americanenglish