#15 Deeptech Analysed - ChatGPT 4: The Latest Breakthrough in Natural Language Processing Technology & Investing in the Future: The 20 Best AI Insights to Watch
What are this week Deeptech business & investment opportunities? 13th March - 20th March
This week…
1️⃣GPT 4 is being released this week
2️⃣20 Best AI insights to invest in
ChatGPT 4: The Latest Breakthrough in Natural Language Processing Technology
By Lirone Samoun , Deeptech Expert
❤️ If you like this format, please like this post and I would definitely make more similar posts in the future.
What is going on?
Unless you live in a cave, you have all heard of ChatGPT. ChatGPT is a language model created by OpenAI that can have conversations with people just like a human would. It's like having a super knowledgeable and helpful friend who can answer any question you have, or just chat with you about your day. You can read more about it on this article.
Why am I talking about GPT3?
Because, ChatGPT rely on the model GPT-3 which is a model that is designed to be able to generate text that sounds like it was written by a human, and it's been trained on a massive amount of data to make that possible.
In fact, it's been trained on over 45 terabytes of text data - that's more than most of us could read in a lifetime!
Thanks to all that training, GPT-3 is able to understand context, mimic human-like conversation, and even write its own creative pieces.
It's already been used for everything from language translation to writing AI-generated news articles, and the possibilities for its applications are truly endless.
If we have seen the amazing possibilities of GPT-3, we would probably be surprised at the power of GPT-4 that is coming this week.
What does it mean?
Microsoft has announced the release of GPT-4 this week.
The biggest feature announced is the fact that GPT4 will be multimodal i.e Not only could you generate text but also video and sound for example.
"We will introduce GPT-4 next week, there we will have multimodal models that will offer completely different possibilities - for example videos," Braun said.
The CTO called LLM a "game changer" because they teach machines to understand natural language, which then understand in a statistical way what was previously only readable and understandable by humans. In the meantime, the technology has come so far that it basically "works in all languages": You can ask a question in German and get an answer in Italian. With multimodality, Microsoft(-OpenAl) will "make the models comprehensive".
You will be able to provide an image, an audio or a video as input and ask for a summary, a description or a transformation.
This means that a new dimension is opening up for designers!
Why does it matter?
💸 For markets: new business opportunities
I don't know about you, but since the arrival of chatGPT, there is not a day that goes by without new applications being released.
Applications that use text understanding, text generation, image generation, audio generation, etc. in many different areas.
Text content summarization, code correction, text comprehension assistance, marketing content generation, semantic Image-to-Photo Translation, streamlined drug discovery and development, personalized medicine, personalized lessons, course design, creation of product descriptions…
We are in the middle of the wave and with the arrival of GPT4 and the big players competing to be the first in the race, new opportunities will continue to present themselves.
What do you want to create?
🧑🏿🤝🧑🏻 For society: Use the tools in smart ways
As discussed in other articles, the arrival of this kind of tools will do a lot of harm on many existing jobs. For example, this media company that cut 12 journalist positions because it was satisfied with the results of ChatGPT.
Moreover, these tools will not necessarily help us humans to be cognitively active.
That's why, in my opinion, we have to use them intelligently and learn how to use them because anyway, we have to adapt to what is going on.
These models can be used by everyone: from the individual who wants to experiment, to the entrepreneur who will build the tools of tomorrow.
We should not be afraid of the progression of AI but rather of the decline of human intelligence.
🔮 What’s next?
While this is excellent news, many people have high expectations and will most likely be disappointed. OpenAI CEO Sam Altman said in an interview that "people are begging to be disappointed, and they will be." when asked about GPT-4.
However, just as we saw the hype around GPT3 and all the developments and applications that were built on it, we can hope that GPT4 will bring a new wave of breath for exponential improvements from big players but also from small players.
It is a new door that will open in the coming days/weeks and new opportunities that will be created.
EDIT: OpenAI just released GPT-4, a large multimodal model! Here's what's new:
- You can use images as inputs and generate captions, classifications, and analyses.
- It can generate/edit creative and technical writing: songs, screenplays, or learn a writing style.
- It can handle 25,000+ words
The api is still not available but people that are subscribed to ChatGPT+ can already play with ChatGPT(4)
Investing in the Future: The 20 Best AI Insights to Watch
By Lirone Samoun , Deeptech Expert
🔎 Latest top news
AI Can read your mind (and maybe soon interpreting your dreams !)
Researchers Used AI and successfully Generated Images Based on People's Brain Activity. They reconstructed high-resolution and highly accurate images from brain activity (measured by FMRI) by using the popular Stable Diffusion image generation model without the need for training or fine-tuning of complex deep generative models
Google AI just announced the PaLM API. It will be released with a new tool called MakerSuite, which lets you prototype ideas, do prompt engineering, synthetic data generation and custom-model tuning
OpenAI releases ChatGPT and Whisper API.
ChatGPT and Whisper models are now available as API, giving developers access to cutting-edge language (not just chat!) and speech-to-text capabilities.Salesforce and OpenAI have launched the ChatGPT app for Slack, which provides instant conversation summaries, research tools, and writing assistance directly in Slack.
Microsoft has launched Copilot for its business apps, with capabilities such as automating repetitive sales and customer service tasks, generating content ideas and assisting with e-commerce product listings.
Salesforce’s Einstein GPT is the first generative AI CRM technology that offers personalized and real-time responses using AI-generated content.
MIT researchers have developed Sybil which can assess a patient's risk of developing lung cancer based on CT scans, achieving strong predictive power.
ChatLLM by Abacus AI: Train a custom chat AI model to remember context and provide accurate responses at a low cost.
MultiDiffusion: An AI model that enables versatile and controllable image generation using a pre-trained text-to-image diffusion model, without finetuning.
Microsoft is releasing Windows 11 update adding its new AI-powered Bing search to the taskbar, alongside other features including Phone Link for iOS, a touch-optimized taskbar, improved widgets, a screen recording feature, and tabs inside Notepad
Amazon recently released a model that outperforms GPT-3.5 by 16% while being 784x smaller.
Have a look on the 2023 MAD (Machine Learning, Artificial Intelligence & Data) Landscape, providing an overview of the current state of the data and AI ecosystem.
💻 CooI AI Tools / Startup
Booth - Create pro quality product photography with AI.
Buildt - AI tool to help developers quickly search and understand large codebases.
Hubble: No-code platform for creating AI-powered applications quickly using advanced AI models like GPT-3 and connecting to data sources
Prime Voice AI: It can clone anyone's voice in seconds
Wonder Dynamic: This AI tool automatically animates, lights, and composes CGI characters into a live-action scene. Without the need for 3D software or production hardware.
Test & Start: Validate your startup idea
Summary.legal - AI-powered transcripts and summaries to help you organize and search video calls and meetings.
Landing AI from Coursera’s co-founder, releases computer vision tool for anyone to use.
PicWonderful - AI-powered photo toolkit
👋That’is for this edition.
❤️Again, If you like this kind of format, please like this post, comment it to show your interest and I would definitely make more similar posts in the future.