- Modern Chaos
- Posts
- MC.11: AI's French Surge, LLM Cost Analysis, WhatsApp's News Frontier, and Bill Gates on AI Agents
MC.11: AI's French Surge, LLM Cost Analysis, WhatsApp's News Frontier, and Bill Gates on AI Agents
This week, we delve into France's rising AI startup scene, the financials of using LLMs, innovative news distribution via WhatsApp, Bill Gates' AI predictions, and much more in the tech world.
Samuel ROY
November 17, 2023
modern chaos issue 11
Modern chaos is a newsletter exploring tech and AI through the journey of a dev agency shifting from services to product design. We share our notes, analysis and experiments.
In this issue:
š«š· Thereās Something Going on with AI Startups in France
š° How Much Does it Cost to Use an LLM?
š± How 13 News Publishers are Using WhatsApp Channels?
š¤ Bill Gates on The Future of Agents
š GPTs and Assistants API: Data Exfiltration and Backdoor Risks in Code Interpreter
š„ļø Open-Interpreter: Self-Improving Feedback Loop
š VimGPT: Browse the Web with GPT-4V and Vimium
š ļø Donāt Build AI Products The Way Everyone Else Is Doing It
šØ Meta: Emu Video and Emu Edit
Tech updates & tools
Thereās Something Going on with AI Startups in France
In just a few months, dozens of French entrepreneurs have shifted their focus to AI, and many companies have raised over ā¬100 million. France is rapidly establishing itself as a key AI hub, driven by startups like Mistral AI and Dust. Their emphasis on regulation and compliance is big enough to set them apart in the global tech landscape. READ IT
How Much Does it Cost to Use an LLM?
Although prices are dropping, using an LLM is still expensive enough to check it impacts your bottom line when integrating the technology into your product. Tomasz Tunguz analyzes the cost of various LLMs for a typical use case. As with other tools, itās important to choose the right one for the job. READ IT
How 13 News Publishers are Using WhatsApp Channels?
NiemanLabs interviews the Financial Times, La NaciĆ³n, The New York Times, Vox, Chilango, the Times of India, and others about their initial experiments with sharing news on the worldās favorite messaging app. It's fascinating to observe how communication evolves. READ IT
š Asking for opinions
Would you like to receive news and stories directly on your WhatsApp?I'm super curious as I haven't heard about it until now. It's a live poll, see the results after voting! |
Bill Gates on The Future of Agents
IBill Gates predicts a radical shift in computing within the next five years, with AI agents offering a seamless, personalized user experience across various domains like healthcare and education. His post underlines the urgency of addressing the ethical and technical challenges inherent in this upcoming technological revolution. READ IT
GPTs and Assistants API: Data Exfiltration and Backdoor Risks in Code Interpreter
Evren presents two scenarios where simple malicious websites target OpenAI Code Interpreter agents and software. These are real threats, and we need to be vigilant about such security risks when designing products with LLMs. READ IT
Open-Interpreter: Self-Improving Feedback Loop
Despite being sped up 4x time, the demo remains impressive. Given a screenshot of the ChatGPT interface, the interpreter recreates the web design, ending up with a good replica. Iād like to see the same tech but in a container to mitigate any security risks on my computer. TWITTER | GITHUB
VimGPT: Browse the Web with GPT-4V and Vimium
The use of LLMs for web browsing is being explored by lots of startups and open-source projects. With this project, Ishan attempts to use only the vision capabilities of GPT-4V for web browsing. GITHUB
Donāt Build AI Products The Way Everyone Else Is Doing It
Builder.io offers an interesting perspective based on their experiments in building a ācopilotā for their product. Their conclusion? Put as little AI as possible into your product to create a better user experience. READ IT
Meta: Emu Video and Emu Edit
Meta continues to advance in the field of generative AI. Emu Edit is worth a look; It feels like image editing will be come a resolved issue sooner than expected. The progress in video generation os also occurring faster than anticipated. META
Reply