Latest news with #Mistral
Android Authority
5 hours ago
- Android Authority
I fired ChatGPT for a week and hired a European AI instead
Nathan Drescher / Android Authority I've cut my ties to a bunch of American tech. Gmail became Fastmail, Keep turned into Obsidian, and I replaced Tasks with ToDoist. I figured AI would be the hardest one to give up. I've been using ChatGPT in an unorthodox way, not for chatting, but as a full-on project management system for scheduling, reminders, and budgeting. Then I stumbled on LeChat a few months ago. LeChat, from the company Mistral, is Europe's answer to the LLM arms race. I used it instead of ChatGPT to run my life for a full week, pushing it way beyond what it was built for. What I found was a mix of blazing speed and blunt honesty, as well as a bunch of limitations to my workflow. Would you consider switching to a non-American AI tool like LeChat? 0 votes Yes, I'm already using a non-American AI tool. NaN % I'm actively looking for alternatives. NaN % Maybe, I hadn't thought of it before. NaN % No, I'm sticking with American tech. NaN % Switching away from American AI Nathan Drescher / Android Authority I've been feeling uneasy about using ChatGPT for a while now. The leadership drama, opaque corporate structure, its close ties to some people in the US government…these already had me questioning my choice to stick with it. But worse than that was how much ChatGPT 4o feels like it is degrading recently. I don't use this LLM for chatting or writing emails. I use it to manage my day, but it was going off the rails by forgetting things I'd input like my diet, giving me randomly hallucinated events when I asked it what was next on my schedule, and other such shenanigans. LeChat felt different. It is built on Mistral's open models and adheres to Europe's strict privacy laws. It doesn't track me outside of the app like Gemini. Plus, its answers are practically instant. So I closed ChatGPT for a week and made LeChat my go-to assistant for everything I'd normally delegate to an AI. This AI is a breath of fresh air Nathan Drescher / Android Authority The first thing I noticed was how ridiculously fast LeChat is at answering. I'd get a full answer for my prompt the moment I hit enter. I thought at first it must surely be missing context or skipping details, but the responses were accurate and useful. This app is set to Ludicrous Speed all the time. It also didn't try to flatter me. During my test, I asked it to schedule a nighttime vampire hunt on my calendar. LeChat told me vampire hunting wasn't real and moved on. When I recreated the same task with ChatGPT, it played along and scheduled it for a Wednesday night. LeChat wasn't rude, just grounded, and that was refreshing. I didn't have to constantly prompt it to keep it short or just answer the question. Web search, calendar, and deep think LeChat also includes a built-in live web search, making it feel a bit like Perplexity. I didn't need to open a special chat or toggle anything. It just searched the web directly inside the thread when needed. I could also turn this off anytime. LeChat connects to Gmail and Google Calendar too, though I didn't test this since I've moved away from those tools. There are also 'Think' and 'Research' modes. The latter is supposedly aimed at summarizing and analyzing source material, possibly competing with Google's NotebookLM. These features aren't in my usual workflow, but I could see them being incredibly useful for students or researchers who need to quickly summarize and analyze large volumes of text. Here's where it couldn't keep up It wasn't all rainbows and unicorns. The biggest problem was memory. LeChat has none. There's no persistent knowledge of who I am or what I've said in other threads. ChatGPT and Gemini can both retain context, so if I ask them to schedule my son's dentist appointment, they remember who he is and even know the name of his dentist. I need to re-enter the information for each and every chat I start with LeChat. I would love to see LeChat incorporate a memory feature that retains key information across sessions. There's no persistent knowledge of who I am or what I've said in other threads. There are no push notifications either. Maybe that works if you're still using Gmail, but I use Fastmail. I couldn't get LeChat to remind me of anything unless I reopened the app and asked it what I had already told it to remember. It also couldn't create documents the way ChatGPT can. I'm talking about those in-line documents that sit outside the main context window, which are especially useful for quick to-do lists or structured notes. LeChat didn't offer anything like that. I understand these tools aren't designed to manage lives, but I've turned ChatGPT into a decent personal assistant. LeChat couldn't keep up. For one-shot answers, web searches, or brainstorming what to feed my picky daughter, it was great. But it fell short for ongoing support. I'm still glad I tried it Nathan Drescher / Android Authority There's something calming about a tool that doesn't try to be everything. Using LeChat made me realize some of what we've grown used to has grown bloated. LeChat is lean, responsive, and no-nonsense. Its speed is mind-boggling for anyone who has grown used to ChatGPT's laborious answers. It's the kind of AI that might appeal to people who don't want AI running their lives, just helping with their day. I'll probably keep using ChatGPT for the heavy lifting. I still haven't found an LLM that can manage projects like ChatGPT. But LeChat has earned a place in my app drawer. Not every tool needs to be a powerhouse. Sometimes it just needs to answer a question and move on, like LeChat.
Time of India
4 days ago
- Business
- Time of India
AI models with systemic risks given pointers on how to comply with EU AI rules
The European Commission set out guidelines on Friday to help AI models it has determined have systemic risks and face tougher obligations to mitigate potential threats comply with European Union artificial intelligence regulation (AI Act). The move aims to counter criticism from some companies about the AI Act and the regulatory burden while providing more clarity to businesses which face fines ranging from 7.5 million euros ($8.7 million) or 1.5% of turnover to 35 million euros or 7% of global turnover for violations. The AI Act, which became law last year, will apply on Aug. 2 for AI models with systemic risks and foundation models such as those made by Google, OpenAI, Meta Platforms, Anthropic and Mistral. Companies have until August 2 next year to comply with the legislation. The Commission defines AI models with systemic risk as those with very advanced computing capabilities that could have a significant impact on public health, safety, fundamental rights or society. The first group of models will have to carry out model evaluations, assess and mitigate risks, conduct adversarial testing, report serious incidents to the Commission and ensure adequate cybersecurity protection against theft and misuse. General-purpose AI (GPAI) or foundation models will be subject to transparency requirements such as drawing up technical documentation, adopt copyright policies and provide detailed summaries about the content used for algorithm training. "With today's guidelines, the Commission supports the smooth and effective application of the AI Act," EU tech chief Henna Virkkunen said in a statement.
The Hindu
4 days ago
- Business
- The Hindu
AI models with systemic risks given pointers on how to comply with EU AI rules
The European Commission set out guidelines on Friday to help AI models it has determined have systemic risks and face tougher obligations to mitigate potential threats comply with European Union artificial intelligence regulation (AI Act). The move aims to counter criticism from some companies about the AI Act and the regulatory burden while providing more clarity to businesses which face fines ranging from 7.5 million euros ($8.7 million) or 1.5% of turnover to 35 million euros or 7% of global turnover for violations. The AI Act, which became law last year, will apply on August 2 for AI models with systemic risks and foundation models such as those made by Google, OpenAI, Meta Platforms, Anthropic and Mistral. Companies have until August 2 next year to comply with the legislation. The Commission defines AI models with systemic risk as those with very advanced computing capabilities that could have a significant impact on public health, safety, fundamental rights or society. The first group of models will have to carry out model evaluations, assess and mitigate risks, conduct adversarial testing, report serious incidents to the Commission and ensure adequate cybersecurity protection against theft and misuse. General-purpose AI (GPAI) or foundation models will be subject to transparency requirements such as drawing up technical documentation, adopt copyright policies and provide detailed summaries about the content used for algorithm training. "With today's guidelines, the Commission supports the smooth and effective application of the AI Act," EU tech chief Henna Virkkunen said in a statement.
Indian Express
4 days ago
- Business
- Indian Express
AI models with systemic risks given pointers on how to comply with EU AI rules
The European Commission set out guidelines on Friday to help AI models it has determined have systemic risks and face tougher obligations to mitigate potential threats comply with European Union artificial intelligence regulation (AI Act). The move aims to counter criticism from some companies about the AI Act and the regulatory burden while providing more clarity to businesses which face fines ranging from 7.5 million euros ($8.7 million) or 1.5% of turnover to 35 million euros or 7% of global turnover for violations. The AI Act, which became law last year, will apply on Aug. 2 for AI models with systemic risks and foundation models such as those made by Google, OpenAI, Meta Platforms, Anthropic and Mistral. Companies have until August 2 next year to comply with the legislation. The Commission defines AI models with systemic risk as those with very advanced computing capabilities that could have a significant impact on public health, safety, fundamental rights or society. The first group of models will have to carry out model evaluations, assess and mitigate risks, conduct adversarial testing, report serious incidents to the Commission and ensure adequate cybersecurity protection against theft and misuse. General-purpose AI (GPAI) or foundation models will be subject to transparency requirements such as drawing up technical documentation, adopt copyright policies and provide detailed summaries about the content used for algorithm training. 'With today's guidelines, the Commission supports the smooth and effective application of the AI Act,' EU tech chief Henna Virkkunen said in a statement. ($1 = 0.8597 euros)
Time of India
5 days ago
- Business
- Time of India
AI models with systemic risks given pointers on how to comply with EU AI rules
The European Commission set out guidelines on Friday to help AI models it has determined have systemic risks and face tougher obligations to mitigate potential threats comply with European Union artificial intelligence regulation (AI Act). The move aims to counter criticism from some companies about the AI Act and the regulatory burden while providing more clarity to businesses which face fines ranging from 7.5 million euros ($8.7 million) or 1.5% of turnover to 35 million euros or 7% of global turnover for violations. The AI Act, which became law last year, will apply on Aug. 2 for AI models with systemic risks and foundation models such as those made by Google, OpenAI, Meta Platforms, Anthropic and Mistral. Companies have until August 2 next year to comply with the legislation. The Commission defines AI models with systemic risk as those with very advanced computing capabilities that could have a significant impact on public health, safety, fundamental rights or society. The first group of models will have to carry out model evaluations, assess and mitigate risks, conduct adversarial testing, report serious incidents to the Commission and ensure adequate cybersecurity protection against theft and misuse. General-purpose AI (GPAI) or foundation models will be subject to transparency requirements such as drawing up technical documentation, adopt copyright policies and provide detailed summaries about the content used for algorithm training. "With today's guidelines, the Commission supports the smooth and effective application of the AI Act," EU tech chief Henna Virkkunen said in a statement.
