The edge of tomorrow - ChatGPT-4 Turbo
OpenAI Developers Conference. Photo by SEJ STAFF, Kristi Hines

The edge of tomorrow - ChatGPT-4 Turbo

The future is approaching us by leaps and bounds. On November 6th Sam Altman, CEO of OpenAI, presented at their first ever developer conference - OpenAI DevDay, although not yet ChatGPT-5, but a fantastic very high-quality update - ChatGPT-4 Turbo.

There are many improvements. First of all, it's a cost reduction for the tokens, which are used for prompts. It is pretty important: ChatGPT is becoming more and more affordable. One more point, which you probably already know, now GPT-4 Turbo’s knowledge cut-off is April 2023, while GPT-4 was trained on web data up to September 2021.

Expanded context window

However, one of the most important updates is the following: GPT-4 Turbo has an expanded context window - four times the size of GPT-4’s. It's the largest context window of any commercial LLM to this moment. Let me explain, it means that now it understands context of your queries better and "keep in mind" more information without losing context. OpenAI says it's “equivalent to more than 300 pages of text in a single prompt.”

For example, GPT-4 Turbo can keep context and summarize: "Rich Dad Poor Dad" by Robert Kiyosaki, "A Brief History of Time" by Stephen Hawking or "Traction: Get a Grip on Your Business" by Gino Wickman. Still the context window is not big enough for “The Intelligent Investor” by Benjamin Graham, which is around 600 pages - we have to read it ourselves :)

Apparently, GPT-4 Turbo users will be able to perform more complex tasks in one prompt. People can even tell GPT-4 Turbo to specifically use the coding language of their choice for results: XML or JSON.

Custom GPTs at OpenAI website

Custom GPTs in GPT Store

In my personal opinion, one of the most exciting news: OpenAI are rolling out the GPT Store later in November, where anybody can publish their own customized versions of GPTs. OpenAI states that you won’t need any coding experience to create your own GPT and share your expertise in the specific field. It means that any specialist in some practical field can build a GPT for a specific purpose.

Moreover, it's also a new business opportunity. I see it as an equivalent of Apple Store or Google Play. Similar to apps in these stores GPTs will be searchable GPT Store, can get reviews, downloads and lead the categories. It's absolutely incredible opportunity and new niche on the AI market: many people will start earning money, depending on how useful and popular their GPT is.

Plus and Enterprise users can start creating GPTs this week at?Chatgpt.com/create. There are some GPTs that are customized for a single purpose, and they are available at OpenAI website already.

Just imagine that instead of looking for a human specialist, you'll be able to hire a GPT acting as experienced Negotiator, Sous Chef, Physics or Literature Tutor. This is an amazing and at the same time frightening prospect. More and more jobs could be replaced by AI. What is your opinion? Write in the comments.
Copyright Shield - OpenAI followed Microsoft. Photo by SEJ STAFF, Kristi Hines

Copyright Shield

The last but not the least, ChatGPT has been protecting OpenAI customers with built-in copyright safeguards. Starting from today OpenAI introduced Copyright Shield — the program which will provide copyright indemnity to enterprise users and pay the costs incurred, if they face legal claims for copyright infringement.

The company stated that they “will now step in and defend our customers, and pay the costs incurred, if you face legal claims around copyright infringement”. Doing this they followed the Microsoft and Google. Earlier Microsoft provided the same protection to enterprise users of Copilot AI, and Google promised to take legal responsibility if users of its generative AI will be sued for copyright infringement.

This is a very important moment for enterprise users and for businesses who are starting to use generative models and AI in everyday practice. The further there will be, the more possible applications of large language models and AI, and the more possible litigation related to copyright violations.

Conclusion

It's becoming obvious that OpenAI's developer conference will be one more important day in IT industry. Specialists and fans will be waiting for OpenAI DevDay as they are waiting Apple presentations.

But the importance of these events is at a completely different level. If new products from Apple, Microsoft or Nvidia are technological or software improvements, then what OpenAI is developing and bringing into the world is much more influential and important. Since it affects both the fate of individuals and entire professions and industries.

Let's hope that the AI singularity (a hypothetical concept where artificial intelligence exceeds human intelligence) will not come soon and we will enjoy the benefits that AI brings, and not curse the day when OpenAI was founded. As the Terminator said "I'll be back", and I promise - with good news.

Sources:

1. OpenAI’s first developer conference, https://devday.openai.com/

2. OpenAI blog, New models and developer products announced at DevDay, https://openai.com/blog/new-models-and-developer-products-announced-at-devday

3. OpenAI blog, Introducing GPTs, https://openai.com/blog/introducing-gpts

4. The Verge, OpenAI turbocharges GPT-4 and makes it cheaper, https://www.theverge.com/2023/11/6/23948426/openai-gpt4-turbo-generative-ai-new-model

要查看或添加评论,请登录

社区洞察

其他会员也浏览了