GPT-4 Turbo OpenAi 核弹级更新，普通人也能做Ai机器人，多模态大整合，Agent和ChatBot开发者亡 #chatgpt

Thx and god bless 😀

做一个培训答疑的AI老师

想做一个关于我自己产品的销售与售后的GPT，能按照我的内容回答客户的问题，并完成销售。

哇。。你突如其来变脸，服装。。已经把我炸了！！🥰🥰🥰

🎯 Key Takeaways for quick navigation:

00:00 🚀 OpenAI GPT-4 Turbo Overview
- OpenAI has released GPT-4 Turbo with 6 major updates.
- These updates will impact a wide range of developers, from those working with PDF documents to speech conversion and more.
- This section introduces the significance of GPT-4 Turbo and its potential impact on developers.
00:58 📜 Context Length Expansion
- GPT-4 Turbo addresses the issue of context length, allowing for much longer texts.
- It can handle up to 300 pages and 128K TOKENs, significantly expanding its capabilities.
- Specific instructions on context length are discussed.
01:58 🧩 Enhanced Control with JSON Documents
- Developers gain more control through JSON documents, enabling the generation of prompt words and other data.
- JSON documents are explained as a tool for generating random prompts.
- The role of JSON in Comfy UI and workflow is highlighted.
02:58 📞 Function Calling and Multiple Functions
- Function Calling is introduced as a feature that allows developers to call multiple functions at once.
- The history of developers seeking this feature and its significance are discussed.
- Mention of third-party development of Function Calling prior to its official release.
03:56 ♻️ Reproducible Output and Data Updates
- GPT-4 Turbo introduces Reproducible Output, enabling consistent answers based on seed numbers.
- Data will no longer be limited to September 2021, with frequent real-time updates.
- The control over AI responses and data updates is emphasized.
04:56 🌐 Multi-Modal Capabilities
- GPT-4 Turbo expands its capabilities to become a multi-modal chat robot, incorporating listening, reading, and writing.
- The comparison is made to an app for visually impaired people.
- Improvements in voice realism and transcribing are discussed.
05:54 🎙️ Whisper V3 and Audio Conversion
- The need for improved accuracy in audio conversion is highlighted.
- Expectations for Whisper V3 with the speed of Whisper Jax and accuracy are expressed.
- Whisper's significance in audio conversion is discussed.
06:53 🛠️ Customization and Fine-Tuning
- Customization through fine-tuning, specifically with GPT4, is introduced.
- The user-friendliness and accuracy of customization are emphasized.
- The potential to train AI with proprietary data is discussed.
07:52 💰 Opportunities in AI Development
- The possibility of profit sharing and building AI platforms is discussed.
- The value of high-quality private data for training AI is emphasized.
- The potential for niche-focused GPT development is mentioned.
08:51 💼 Real-World Applications
- Real-world applications of custom GPTs are highlighted, including programming courses, design, and scheduling.
- The integration with tools like Zapier is discussed.
- The potential for AI to handle various aspects of daily life is mentioned.
09:50 🤖 Building Personalized GPTs
- The process of building personalized GPTs is described, emphasizing its simplicity.
- The idea of becoming a GPT internet celebrity is introduced.
- The upcoming content related to developers in the conference is mentioned.
10:50 📊 OpenAI's Forward-Thinking Approach
- OpenAI's commitment to addressing AI and Chatbot challenges is discussed.
- The integration of modalities and the convenience of a unified AI are emphasized.
- The pace of AI development and the competitive landscape are briefly touched upon.
11:49 📺 Video Conclusion and Audience Engagement
- The video concludes by encouraging viewers to engage by leaving comments.
- A related video is mentioned for viewers to explore.
- The potential for creating new ideas and sparks is suggested.

Made with HARPA AI

我想做一個出一張嘴定好規則，就可以幫我生成遊戲app 還可以同時提供跨平台版本。

根据我浅薄的互联网经验，任何高级技术高成本技术，能迅速盈利的方法肯定都包括一点，那就是搞色情😂

我想用GPTS.可以告诉我用什么版本电脑最好❤

真的强

最後的 grant 500 usd 好爽，真的被這發表會驚艷到，比 Apple 那些都還扯。

做小秘和做机器人有什么区别吗

怎么订阅JOJO的课程啊

为什么回复问题就卡住了？

主播的用词好有趣。。。。另外，说事实比较安全。后面的评论很容易被打脸。

老师好！我是你的粉，你能推荐一下在哪里下载open吗？我们下载的好多都不怎么好用！麻烦推荐一下

,太複雜了

赚的收益，在哪里提现，或者是发布后的商店机器人，我的提现账户如何知道在哪里

这一期，美人说的很关键，如果是真的自己训练出来的可以卖钱复用并分佣，这是划时代的模式，比大语言模型本身更有价值

up主没有必要过度美颜吧😅

open AI 這種做法....是想統包所有的東西, interface 還是要留著不然就不是很好

目前工具的複雜程度，與早年要開發遊戲的困難度類似，當年連個 2D, 3D Engine 都要自己搞，甚至連 Engine 這個字都還未曾在遊戲領域出現，結果現在可好了，各種花式引擎隨便一個學生拉幾個資源進去，就做出來以前大師要花半年甚至兩、三年才能作出來的東西。相信再過不久，也會出現所謂 AI Engine 這樣的東西，整個 AI 功能由一個核心工具和環境驅動，那時候要搞起 AI 物件，就不會像現在 99 這麼辛苦了。

GPT-4 Turbo如何使用？听说是随机给名额？

Ai进展的程度，情况就像当年PC推出，有64K已经非常不错，如果您有256KB那就令人十分羡慕，如果您的PC是512KB那是有钱人了。PC的世界需要好几年才能让512KB为标准装配，可是AI的进展更令人瞠目，相对来说，就几个月时间而已，已经超越PC的几年的进展。

Jojo , 有沒有收費咨詢的服務？