• The Neuron
  • Posts
  • 😺 Honest review of ChatGPT Voice Mode

😺 Honest review of ChatGPT Voice Mode

PLUS: A "striking" similarity between ScarJo and ChatGPT

Welcome, humans.

Our readers are like AI guinea pigs: always hungry to test the latest AI tools and products.

So if you’re working on AI, consider sponsoring The Neuron to get your brand in front of 450,000+ AI enthusiasts. Fill out this form ASAP…spots are quickly filling up!

Here’s what you need to know about AI today:

  • We give our honest reaction to ChatGPT Voice Mode.

  • Apple’s upcoming AI-powered Siri will be able to take over your apps.

  • Suno v3.5 now creates full 4-minute songs from your text prompts!

  • Nvidia released even more AI chips.

ChatGPT Voice Mode: impressive, though not flawless.

We've spent the past week playing with ChatGPT Voice Mode, the newest iteration of ChatGPT-4o that you can speak to.

If you haven’t caught a glimpse of what it’s capable of yet, go read our write-up of Voice Mode from when GPT-4o was launched—it certainly feels like a glimpse into the future.

Here’s our honest reaction: It’s the first time we feel like we’re having a true conversation with AI, not a back-and-forth riddled with 3-second pauses and “Sorry, could you say that again?”

The real game-changer is the speed. Ask Voice Mode any question, like “Tell me about neural networks,” and it typically replies in under a second. This is a big improvement.

Here’s when we’ve been using Voice Mode:

  1. When we want to interact with ChatGPT, but prefer to speak over type (e.g. dictating notes, brainstorming sessions, or learning new topics).

  2. When we’d normally use Siri or Google.

For instance, yesterday we had a convo with ChatGPT about the differences between Hatha and Vinyasa yoga. Whenever we had additional questions or needed more details, we just hit the 'Tap to Interrupt' button.

Of course, Voice Mode isn't without its quirks.

For every 5ish questions, Voice Mode mishears what we said and then veers off on a tangent. When we asked about aliens, it responded by talking about Alienware instead. Obviously, a computer would prefer talking about Alienware…

It also interrupts (OpenAI really did train it to mimic humans!). Any time we pause to think and formulate our next sentence, ChatGPT often assumes we're finished and begins to respond.

And yes, it hallucinates like the ‘ol ChatGPT. When we inquired about who won a tennis match at Roland Garros on Saturday, it gave the correct winner but quoted the wrong score.

Why it matters: For now, ChatGPT Voice Mode is mostly a fun toy. But as more users get their hands on it, we’ll see it applied in both productive ways and, naturally, ways that make our brains hurt.

Some users with Voice Mode access jumped straight to using it as a substitute boyfriend, and at this point, are we really shocked?

Related reading: Voice analysis shows striking similarity between Scarlett Johansson and ChatGPT (link).


Automate critical business tasks with Cohere on AWS.

Here’s a little secret for those of you who are keen to automate critical business tasks but are worried about privacy and security.

Use Cohere’s models on AWS (check out the deets here).

Cohere lets you do handle everything you do with ChatGPT—copywriting, summarization, and data analysis—while ensuring your data remains private.

And this webinar from AWS will show you exactly how to do it, step by step.

Bonus: They’ll also demonstrate how to tailor Cohere’s models and leverage RAG to build AI applications.

Around the Horn.

  • Apple’s upcoming Siri update will let you control your iPhone apps with just your voice.

  • AI adoption in at least 1 business function has risen from 33% of companies to 65% in the last year.

  • Nvidia announced new AI chips…AGAIN!

Treats To Try.

  1. *Brilliant offers bite-sized AI lessons to help you stay competitive at work. Get 20% off a premium annual subscription for life by using this link before it disappears!

  2. ElevenLabs Sound Effects lets you create any sound imaginable from text descriptions.

  3. NOVA-1 from Higgsfield is an AI video generator that lets marketers craft social media content using their brand’s brand assets.

  4. Suno v3.5 can now generate 4-minute songs and additional 2-minute song extensions from your text prompts. Plus, Suno is paying $1M to the top Suno Creators this year!

*This is sponsored content. Advertise in The Neuron here.

Monday Meme.

A Cat's Commentary.

That’s all for today, for more AI treats, check out our website.

Get your brand in front of 450,000+ professionals here.

See you cool cats on Twitter: @nonmayorpete & @noahedelman02

What'd you think of today's email?

Login or Subscribe to participate in polls.