- AI Tool Report
- Posts
- ✨ New Claude beats GPT-4o
✨ New Claude beats GPT-4o
🛡️ OpenAI reveals cyberdefense breakthroughs
Martin Crowley & Liam Lawson
June 21, 2024
Welcome to AI Tool Report!
Friday’s top story: Anthropic has released its newest model—Claude 3.5 Sonnet—and says it outperforms OpenAI’s GPT-4o, Google’s Gemini, and Meta’s Llama…
🌤️ This Morning on AI Tool Report
✨ New Claude beats GPT-4o
🎓 How to learn the concepts behind AI
🛡️ OpenAI reveals cyberdefense breakthroughs
🤑 How to demand 77% more compensation with AI
📝 How to create content with ChatGPT
🏙️ Paris expands AI start-up hub
🎙️ Voiceover artists facing extinction?
Read Time: 5 minutes
CHATBOTS
Our Report: Anthropic has revealed its newest AI model—Claude 3.5 Sonnet—which reportedly performs better across a variety of tasks than OpenAI’s GPT-4o and Google’s Gemini, and is available to web and iOS Claude users.
🔑 Key Points:
It’s twice the speed of its predecessor (Claude 3 Sonnet), outperforms its biggest model (Opus), understands more complex instructions, interprets charts accurately, and transcribes text from “imperfect” images.
It also outperformed OpenAI’s GPT-4o, Google’s Gemini 1.5 Pro, and Meta’s Llama 3 in 7/9 industry benchmarks and 4/5 vision benchmarks, as it was trained on new and AI-generated data.
Claude 3.5 Sonnet also drives a new feature (called Artifacts) which allows users to edit and add to content—like code, emails, or documents—that have been generated by Claude, within the app.
🤔 Why you should care: Anthropic just launched Claude 3 Opus in March, and its newest release comes hot on the heels of OpenAI’s GPt-4o and Google’s Gemini 1.5 Pro, all launched within months of each other, highlighting the rapid pace of innovation, improvement, and competition within the industry. However, recent improvements to these models have been incremental: There hasn’t been a massive leap of progress since the upgrade of GPT-3 to GPT-4, possibly because of the amount of compute needed to train these models.
Which of these leading AI models–Claude, Gemini, Llama, Copilot, and GPT–do you prefer? |
Together with Brilliant
How do you stay ahead of the AI curve? You understand the concepts powering it.
Brilliant makes it easy—with a huge library of quick, interactive lessons in AI, programming, logic, data science, and more. From LLMs to neural networks and beyond, Brilliant takes a first-principles approach to learning.
You’ll master core building blocks before moving on to more advanced skills. Dive into:
Bite-sized, hands-on lessons that make complex AI concepts accessible and engaging
Personalized features and competitive elements to keep you motivated and on track
Hands-on problem-solving and real-world applications for impactful learning
Join over 10M people and try it for 30-days, at no-cost. Plus, AI Tool Report readers can snag a concession on an annual premium subscription.
CYBERSECURITY
Our Report: Following the launch of its Cybersecurity Grant Program in 2023—designed to equip cybersecurity defenders with advanced AI models and support research into how AI can be used in cybersecurity—OpenAI has shared some of the groundbreaking projects they have supported during this program.
🔑 Key Points:
Since the program launched, OpenAI has supported various cybersecurity projects including protecting large language models (LLMs) from prompt-injection attacks and cybersecurity threats.
It’s also sponsored projects using AI to automate the detection and reduction of software misconfiguration (a common cause of security incidents), as current methods rely on outdated rules-based policies.
Plus, it’s working with labs to build AI-powered defense techniques that prevent attacks (without compromising accuracy or efficiency) and improve the ability of LLMS to identify and fix code vulnerabilities.
🤔 Why you should care: OpenAI received over 600 applications for the program, highlighting the critical need and potential to improve current cybersecurity defenses and, to show its support for the cause, has given ChatGPT Plus access to many in the cybersecurity space (at no cost), seeing it “as key to enhancing AI adoption in cyber defense.”
In Partnership with AI For Professionals
2 months ago, we released the first cohort of AI For Professionals 👇
Led by AI industry professionals with over $100M+ in funding
1000+ students already enrolled
4+ hours of premium content
We’ve added 4 new modules:
How to build a 7-figure business using an AI copilot
10x your productivity with smart prompts & custom GPT’s
Your comprehensive introduction to AI
Write with AI & smash it on LinkedIn
Ready to take your AI education to the next level and demand 77% more compensation?
Friday’s prompt: How to create content with ChatGPT
Type this prompt into ChatGPT:
“Write a [type of content] using the Rule of One framework that focuses on one main [idea], one main [message], or one main [CTA] to make the content more effective and memorable.”
Results: After typing this prompt, you will get an effective and memorable piece of content that follows the ‘Rule of One’ framework.
P.S. Use the Prompt Engineer GPT by AI Tool report to 10x your prompts.
START-UPS
🏙️ Paris expands AI start-up hub
Our Report: Paris-based AI start-up, Poolside—which is building an AI coding assistant, designed to speed up software development—is expecting to raise over $400M, giving it a $2B valuation, post-funding, cementing Paris as a hub for AI start-ups.
🔑 Key Points:
It previously raised $126M last August from multiple backers, including Point Nine, Redpoint, and Bain Capital Ventures—which are rumored to be co-leading this funding round.
Investors are impressed with the founding team's experience, as CEO Warner was the former GitHub CTO, and CTO Eiso Kant, founded the AI dev tool firm, Athenian, which was acquired by Linux.
The coding assistant (not yet publicly available) will just focus on one objective: To help developers work quicker, which is unlike many existing models that have a more generalized approach.
🤔 Why you should care: This latest funding round is further bolstering Paris as a major hub for AI software development start-ups, with the likes of Mistral and H (which have raised $113M and $220M, respectively) also emerging from the ‘City of Lights’, but it’s raising concerns over market saturation: Will Poolsides experience and singular focus on developer productivity equate to success in such a crowded market?
AUDIO
🎙️ Voiceover artists facing extinction?
Our Report: Audio platform (Pocket FM) has partnered with voice-cloning company (ElevenLabs) to convert text (such as scripts) into audio, using AI, potentially leaving voiceover artists out of work.
🔑 Key Points:
Pocket FM was already using ElevenLabs to try and convert text content into audio, and has now decided to expand the partnership and make the tool available to help writers convert their work into audio series.
During testing, Pocket FM produced over 30,000 hours of audio—while slashing audio production budgets by 90%—and expects to triple its content library (which already has 100,000 hours of audio) this year.
Co-founder Prateek Dixit believes that, using current methods, writers can just produce 30mins of high-quality audio a day, but with this new AI tool, they’ll be able to produce 10X more.
🤔 Why you should care: Like many in the creative industry, the role of the voiceover artist could face extinction if the use of AI is not properly regulated, as there are already reports that voiceover artists' voices are being used for training, without consent, leaving them with no livelihood.
We read your emails, comments, and poll replies daily.
Hit reply and tell us what you want more of!
Until next time, Martin & Liam.
P.S. Don’t forget, you can unsubscribe if you don’t want us to land in your inbox anymore.
What did you think of this edition? |