Hacking ChatGPT Advanced Voice Mode to Make it 10x Smarter

preview_player
Показать описание

In this video, Mike Roberts, founder and CEO of SpyFu and RivalFlow, dives into an impromptu test of the new Advanced Voice Mode in ChatGPT.

Mike pushes boundaries by trying to run a full Q&A podcast with the AI, even getting creative by having Microsoft Word read context aloud to ChatGPT.

It’s a rollercoaster ride as he explores whether conversational AI can handle real, dynamic back-and-forth dialogue, with plenty of surprises—like almost knocking over the camera! 😂

Throughout the conversation, Mike also shares insights about RivalFlow, a tool designed to support content creators in a way that's effective and ethical, without the shady tricks often associated with SEO.

He talks about how SEO is evolving, the role of AI in that evolution, and the value RivalFlow aims to bring.

🚀 Topics Covered:
- Testing ChatGPT's new Advanced Voice Mode
- Challenges of keeping AI in context during a live conversation
- The journey behind building RivalFlow for modern SEO
- Insights into the future of SEO and AI's growing influence

👉 Curious about the future of conversational AI or the ethical side of SEO? This experiment offers a unique, candid look at how far AI has come... with a few hiccups along the way.

Connect with us on:

Chapter Markers:
00:00 - Introduction
00:09 - Advanced Voice Mode in the ChatGPT App
00:36 - Exploring the Differences Between Modes
01:00 - Trying to Download the Latest Version & Expected Features
01:21 - Switching Between Advanced and Basic Modes
01:57 - Experiment with Feeding ChatGPT Context via Microsoft Word
03:03 - Mike Roberts Introduces Himself and RivalFlow
04:15 - Reflecting on Using Microsoft Word for Context Feeding
05:00 - Creating SEO Workflows that Do No Harm
06:09 - Google Search Console Integration and Seeing Results
06:54 - Mike’s Purpose: Adding Value Through Software
07:51 - Evolution of SEO and Google’s AI Integration
08:45 - RivalFlow’s Approach to User-Focused Content
09:27 - Content Creation Strategy with AI Augmentation
10:06 - Rejecting AI Content for Ranking's Sake
10:57 - RivalFlow 2.0 Updates and Improvements
12:00 - Competitor Analysis and Identifying Gaps
13:00 - Blending Content to Fit Users’ Voice
14:18 - Further Evolution of RivalFlow and AI Capabilities
15:24 - Mike’s Impressions on ChatGPT’s Context Handling
16:03 - Exploring Future RivalFlow Features
17:15 - Holistic Content Analysis with RivalFlow
18:00 - AI-Driven Content Strategy and User Needs
18:54 - Mike Reflects on AI and Human Creativity
19:48 - Final Remarks & Posting the Video
Рекомендации по теме
Комментарии
Автор

Wow, this was better than a podcast! The MS word context was a great idea.

michaelharalson
Автор

That’s phenomenal Mike. It just occurred to me that you could have asked the Voice Model to revise your Word doc with extra ideas gleaned during “your” interview!

riothecat
Автор

7 minutes in- you don't need AI to chat, you've not given it any chance to say anything 😂

HeartofthePiano
Автор

My favorite part was when it went into 2x when I was already in 2x and got to listen to the word doc in 4x.

TehLissness
Автор

Hmm. I think i was able to give it context when pasted after I speak with it beforehand. Did I miss something?

SavesTheGame
Автор

I ran into the same problem you did. I was able to paste the text into base 4o, have it 'remember' it, then have Voice recall that info.

Corbald
Автор

You are a divergent thinker. I subscribed! Great job!

WorldCollections
Автор

Well, you can also paste the word into the custum instructions

metanulski
Автор

You can feed it context in the system prompt, go into settings and click personalization

PrinceCyborg
Автор

It comes over as using the ChatGPT Advanced Voice mode hype to plug your products 😅🤣

Dutchruler
Автор

The model is built off GPT-4o, so it's not going to be able to start improvising with you on the subject of itself, except to the extent that OpenAI has provided some new details in the underlying system message

jason_v
Автор

Interesting, chat with AI is awesome!

coola
Автор

What you need to do if you want it to have context is to put them on your custom instructions😅, I think that's about 800 character limits

adolphgracius
Автор

Total self promotional bs. 8 minutes in and nothing to do with the title of the video.

PromptTheory
Автор

You can't use any other multi modal features with AV. It's very limited and pretty much nothing more than a parlour trick at the present time.

cbnewham_ai
Автор

Surely if you have a voice chat within a thread in which you already have context, ChatGPT will be able to converse with knowledge?

I didn’t understand why, even after giving it the lengthy context using MS word, you still have in-depth answers when having the ‘conversation’?

Oh, and your notifications on your phone were annoying.

MartinKoss
Автор

TLDR: read outloud (or use word text to speach) the context to get better context delivery. this doesn't make chatGPT smarter.

Toxicflu