✅ Pitched Giskard (Security for #LLM agents) to 2 French ministers... A bit stressful but check 😄

I had a great time at the #AISummit Business Day in Paris, meeting an impressive variety (> 4000 people in #stationf!) of politicians, entrepreneurs, researchers and enterprise AI leaders.

The AI ecosystem is vibrant, and France is playing the locomotive role for the EU to catchup with the US and China!

It's just the beginning, we have much to prove and deliver.

Happy New Year, everyone!

I’m back from a quick social media break with fresh news from the Giskard team. I’m thrilled to share our 2-minute video highlighting last year’s biggest achievements and what we’re gearing up to deliver in 2025.

Don’t miss it - watch now and tell me what you think!

Excited to launch a special #giveaway contest for the Giskard #community! 🎁
It's our way of saying thanks for your amazing support ahead of holidays.

🌟 To enter:
1. Like and share this post.
2. Join our Discord: https://gisk.ar/4fVGEHf
3. Fill out this quick form: https://gisk.ar/49lss7I

The contest will end on December 10 and winners will be announced on December 11. Join the fun, win big!

Good luck! 🎉

#AIGiveaway #CommunityLove #WinWithUs

Overall, I found this essay a well-written piece, and I appreciate someone like the CEO of Anthropic dedicating time to write his thoughts on these important topics.

To read the essay yourself: https://gisk.ar/4fdR4kG

𝑪𝒐𝒏𝒄𝒍𝒖𝒔𝒊𝒐𝒏
I would have preferred less grandiose statements (e.g., "transcendent") as the author had stated in the intro he meant to "avoid grandiosity," and more focus on questions such as:
* What's the strategy to better distribute the benefits of AI within society and across countries?
* How to make sure AI benefits the education system and vice versa, how to educate people to use AI?
* How to combat ill usages of AI?

𝟱. 𝗪𝗼𝗿𝗸 & 𝗠𝗲𝗮𝗻𝗶𝗻𝗴 🤔
This section makes interesting points about AI automation leading to more leisure time and how human economies might adapt to a potentially post-work world, with optimism drawn from past historical economic transitions.

𝟰. 𝗣𝗲𝗮𝗰𝗲 & 𝗚𝗼𝘃𝗲𝗿𝗻𝗮𝗻𝗰𝗲 ✌️
Like Dario, I "see no strong reason to believe AI will [...] structurally advance democracy and peace." The relationship between AI, peace & democracy is complex, with both positive implications (information access) and negative ones (mass surveillance). I regret the relative absence of discussion on AI regulation.

𝟯. 𝗘𝗰𝗼𝗻𝗼𝗺𝗶𝗰 𝗗𝗲𝘃𝗲𝗹𝗼𝗽𝗺𝗲𝗻𝘁 📈
This is where I expect the most from AI, as it can make access to education & knowledge far easier, lowering barriers of entry to many jobs. Unfortunately, Dario didn't mention it, focusing again on health. Besides, the article doesn't address inequality or redistribution challenges, especially given high AI infrastructure costs.

𝟮. 𝗡𝗲𝘂𝗿𝗼𝘀𝗰𝗶𝗲𝗻𝗰𝗲 & 𝗠𝗶𝗻𝗱 🧠
Dario makes similar claims as Biology & Health, focusing on AI helping create new cures for psychological issues. My issue: I find it troubling that mental health is only viewed through the lens of new drugs.

𝑲𝒆𝒚 𝑨𝒓𝒆𝒂𝒔 𝒐𝒇 𝑰𝒎𝒑𝒂𝒄𝒕

𝟭. 𝗕𝗶𝗼𝗹𝗼𝗴𝘆 & 𝗛𝗲𝗮𝗹𝘁𝗵 🧬
The prospect of AI discovering new medicines & curing diseases is definitely the most exciting and positive application. I appreciate Dario's intellectual honesty in acknowledging that while AI can accelerate discovery processes, there are still physical limits related to conducting real experiments.

𝑰𝒏𝒕𝒓𝒐𝒅𝒖𝒄𝒕𝒊𝒐𝒏
I also believe we should work on a positive scenario of AI adoption. #AI can have real upsides for our society, and companies focusing on AI risks (like @Giskard) are not AI "doomerists" - we're trying to make that positive AI scenario happen.

The framework chosen is correct in my opinion - it's more helpful to talk about "powerful AI" than "AGI". "Powerful" AI invites reflection about what will be AI "powers" in the next 10 years. The notion of power is multifaceted.

I finally completed reading the essay "Machines of Loving Grace" by Anthropic CEO Dario Amodei. Here's my personal review. 📚

The skeleton key #AI #jailbreak

https://www.perplexity.ai/page/the-skeleton-key-ai-jailbreak-OuIr1gvxRQO0O2Bu6ZBI1Q#48b9ab5a-2dbe-4150-ab4c-0243512819c1

PortSwigger, the company behind the Burp Suite of security testing tools founded by security expert "Daf", raised $112M, its first outside investment (Ingrid Lunden/TechCrunch)

https://techcrunch.com/2024/06/27/portswigger-the-company-behind-the-burp-suite-of-security-testing-tools-swallows-112m/
http://www.techmeme.com/240627/p19#a240627p19

I play a lot with making AI, specifically GPT, sound more human. I started with experimenting with writing fiction with AI. As anyone who's tried that can tell you: it doesn't really work. It can be done, but it's more technical hobby (can I get the AI to do XYZ?) than writing method. And real #writing is more fun anyway.

Point is, the #artists are right, even good #AI writing is soulless. If you only write short SEO content you might have some real AI competition, but most writers? Cheer up!

Interesting #science news about the extension of the periodic table ⤵️

https://www.perplexity.ai/search/Extension-of-the-8HsNR1ttTxCCWVU59tTUZA

The story of Superfest, the ancestor of... Gorilla Glass 🥃

https://www.youtube.com/watch?v=vEvBpjCOBu0

Mistral AI launches Mixtral 8x22B, its latest sparse mixture-of-experts model, after releasing Mixtral 8x7B in December 2023 (Shubham Sharma/VentureBeat)

https://venturebeat.com/ai/mistral-ai-drops-new-mixture-of-experts-model-with-a-torrent-link/
http://www.techmeme.com/240410/p33#a240410p33

Legendary Valve founder Gabe Newell launches new brain computer startup 🧠👀

https://www.perplexity.ai/search/Valves-founder-braincomputer-ULXvhovpRpqTE3rJLgbYgA