Google just dropped its new Gemini Live feature today, but before it got announced, I got a chance to experience Google’s new ChatGPT Voice Chat competitor for a short spell — and it’s arguably the biggest threat to OpenAI’s popular AI tool.
I wouldn’t quite say that Gemini Live beats ChatGPT just yet. After all, I didn’t get enough time with Gemini Live to say that for sure. However, from what I saw at a recent press briefing with Google, OpenAI should be shaking in its boots.
OpenAI has a ‘99% effective’ ChatGPT-detection tool ready. So why aren’t they releasing it?
Gemini Live can now have continuous conversations
As a Gemini Advanced subscriber, the premium version of Google’s AI chatbot, I couldn’t quite have a continuous voice conversation like I could with ChatGPT — until Gemini Live came along.
Credit: Kimberly Gedeon / Mashable
For example, with ChatGPT, listening mode can remain on as long as I’d like, allowing ChatGPT to converse with me without needing constant input, seamlessly responding to ongoing spoken interactions.
With Gemini Advanced, prior to Gemini Live, I couldn’t do the same. I’d have to tap the microphone button to get it to “listen.” Once I said my peace, listening mode would end in just a few seconds before Gemini jumped in with a response. This interaction is similar to what you’d experience with Siri or Google Assistant. You invoke it; it listens; you speak; it responds; and the conversation stops — that’s it. There was no uninterrupted conversation capability.
Credit: Kimberly Gedeon / Mashable
However, with Gemini Live, I got to experience a new-and-improved Google AI chatbot. Similar to ChatGPT, I was able to choose a preferred voice from 10 options. I rejected voices like “Lyra,” “Ursa,” and “Capella” because they were too bubbly for my tastes. In the end, I opted for “Orbit.” Why? Nothing more than I thought he sounded attractive. (What? I want my AI chatbot to sound hot.)
After selecting Orbit as my official Gemini Live voice, a Google rep hit the waveform icon at the bottom-right of the Gemini app (on one of the Pixel 9 phones that dropped today). Next, a glowy UI appeared, indicating that Gemini is listening and ready to have a conversation. Here’s what was said to kick off the interaction: “I’m looking to become more of a morning person. What are some ways I can get started?”
“That’s a great question!” Gemini said. “To start, try setting a regular sleep schedule. Go to bed and wake up around the same time each day, even on weekends. It helps regulate your body clock. Also, try getting some bright light as soon as you wake up. It helps signal your body that it’s time to be alert.”
Gemini Live has contextual continuity
Gemini Live also appeared to have great context retention. In other words, the chatbot was capable of building upon the context of our conversation based on our previous interaction.
Credit: Kimberly Gedeon / Mashable
For example, “And what about caffeine?” was the next question directed at the Google chatbot. Gemini continued the dialogue in a coherent manner, adding, “Caffeine can definitely help give you a morning boost, but if you’re sensitive to it, try limiting your intakes earlier in the day. Also, remember caffeine’s got a long half-life, so even a late afternoon coffee will keep you up at night.”
In response to “And what about working out?”, Gemini said, “Working out in the morning in the morning can make you feel more alert.”
Gemini Live can be interrupted
On the current iteration of ChatGPT that’s available to the general public, you can interrupt ChatGPT, but it’d require you to tap a button. On Gemini Live, we interrupted the Google AI chatbot with a question about sleep, sans tapping on any buttons, while it delved deeper into how exercising can aid with one’s “morning person” goals.
Credit: Kimberly Gedeon / Mashable
It trailed on for a bit, but it suddenly stopped mid-sentence before adding, “To improve your sleep, try to have a consistent bedtime routine. Maybe read a book. Take a warm bath. Or do some light stretching.”
Again, we didn’t need to press anything to interrupt it.
Credit: Kimberly Gedeon / Mashable
Similar to ChatGPT, you can find a transcript of your chat once you’re done speaking to Gemini Live.
Gemini Live will be able to “see” things soon
At the Made by Google event, it was revealed that Gemini Live will be able to “see” soon. Using your camera, you can, for example, point at math equations on a paper for homework help.
Credit: Kimberly Gedeon / Mashable
You can also have it “look” at your bedroom to get inspiration on how you can spruce it up. Gemini Live will also become a better researcher, offering multi-step plans on how to achieve your goals. For example, let’s say you want to open a restaurant business in New York City, but you have no idea how to even start. Instead of taking hours to research the steps you need to take, Gemini Live will outline a step-by-step plan for you, allowing you to figuratively get to where you want to go faster.
How to get Gemini Live
Gemini Live did a lot of catching up with ChatGPT. Gemini was behind when it came to voice capabilities, but this changes with Gemini Live, making it a true threat to OpenAI’s chatbot tool.
Curious about how you can get Gemini Live? You must be a Gemini Advanced subscriber to get it. It officially drops on Tuesday, Aug. 13.
