Google Revolutionises Biology ๐งฌ Prompt Engineering is Dead ๐AI Adoption Surges ๐
Your weekly AI wrap
10th Edition of This Week in AI
Welcome to the 10th edition of this newsletter, thank you for taking the time each week to read my musings on the rapid evolution of AI.
Also, due to a scheduling error this edition is coming out on a Friday instead of the usual Thursday.
Letโs get into it!
We can say right now, with a high degree of scientific certainty, GPT-5 is going to be a lot smarter than GPT-4.
GPT-6 is going to be a lot smarter than GPT-5. And we are not near the top of this curve and we kind of know what to do. And this is not like it's gonna get better in one area.
It's not that it's always gonna get better at this eval, or this subject, or this modality. It's just going to be smarter in the general sense. And I think the gravity of that statement is still like underrated.
OpenAI CEO Sam Atlman at talk at Stanford last week.
Google's AlphaFold 3 Revolutionises Biology with Unprecedented Structural Predictions
Google DeepMind and Isomorphic Labs have unveiled AlphaFold 3, an AI model that accurately predicts the structure and interactions of a vast range of biological molecules, including proteins, DNA, RNA, and ligands.
This breakthrough has the potential to transform our understanding of life processes and accelerate drug discovery.
Why it matters: AlphaFold 3 represents a massive leap forward in AI-powered biology. It offers scientists unprecedented insights into how molecules interact, paving the way for advancements in multiple fields:
Understanding Life: AlphaFold 3 helps researchers visualize complex cellular systems, potentially leading to breakthroughs in disease research and the development of new treatments.
Drug Design: The model's ability to predict protein interactions with drug-like molecules opens doors to the design of new medications, even for previously untreatable diseases. (AlphaFold 3 has surpassed existing methods in accuracy by at least 50%).
Resilient Crops and More: AlphaFold 3's applications extend to developing biorenewable materials and improving crop resilience.
Democratising Research: The free AlphaFold Server provides scientists globally with accessible and powerful tools, regardless of their resources.
๐ Great summary (easy to understand)
Microsoft Debuts 'Air-Gapped' AI for Intelligence Agencies
Microsoft has created a secure AI model tailored for the U.S. intelligence community to process classified information. This model, based on GPT-4 technology, operates entirely offline in an "air-gapped" environment to ensure data protection.
Why it matters: This breakthrough represents a significant step for AI adoption in sensitive government work. The isolated nature of the model addresses security concerns that have previously hindered the use of generative AI for classified information.
It highlights Microsoft's commitment to developing AI tools for government agencies, even amidst recent criticism about its cybersecurity practices.
๐ฐ Article by Quartz
The Uncertain Future of Prompt Engineering: AI May Do It Better
Prompt engineering, the art of crafting prompts for AI tools, is becoming less reliant on human intuition.
New research shows that AI models can automatically generate prompts that outperform those created by experts. This raises questions about the longevity of prompt-engineering jobs.
Why it matters: This development has significant implications for businesses and individuals using AI:
Improved AI Performance: Auto-tuned prompts lead to better results from AI tools, potentially boosting productivity and generating more creative outputs.
Evolving Skillset: Prompt engineering may become less about finding the "perfect phrase" and more about understanding AI models and designing effective evaluation systems (LLMOps).
AI Self-Optimization: As AI models get better at prompting themselves, human intervention may become less necessary, streamlining the process.
Key Points:
Bizarre AI Prompts: Auto-generated prompts are often strange and counterintuitive, highlighting the difference between AI and human "thinking" (e.g., Star Trek references boosting math scores).
Automation Triumphs: Auto-tuning consistently outperforms human prompt design, particularly for tasks like image generation.
Shifting Job Landscape: While roles focused on AI adaptation will remain crucial, the specific title of "prompt engineer" could become less common.
๐ฐ Article by Spectrum
AI Adoption Surges: Employees Demand AI Tools, Outpacing Companies
Microsoft and LinkedIn's report reveals a staggering rise in workplace AI use. Employees are bringing their own AI tools to increase efficiency, even as many companies struggle to develop AI implementation plans.
Why it matters: This trend signals a major shift in workplace dynamics. Companies must urgently adapt to provide AI solutions or risk losing out to competitors and a workforce that's already adopting AI independently (78%).
The report emphasizes the growing importance of AI skills for job seekers (66% of leaders prioritise AI skills). Those who embrace AI become "power users," gaining significant advantages in creativity, focus, and job satisfaction.
๐ 2024 Work Trend Index Annual Report (PDF)
AI Trends in 2023: Open-Source Models Surge, US Dominates, and Regulation Takes Off
Stanford's 2023 AI Index reveals a shift towards open-source foundation models (used in tools like ChatGPT). The US leads in AI development and investment, while generative AI attracts significant private funding.
Although AI adoption is increasing, concerns about job displacement and AI-powered products are also on the rise.
Why it matters: This comprehensive report highlights key trends shaping the future of AI, offering valuable insights for businesses and individuals:
Open-Source Growth: While closed-source models still have a performance edge, the rise of open-source AI signals potential for broader innovation (65.7% of new models).
US Dominance: The US cements its position as the undisputed leader in AI (models, investment), prompting questions about global competitiveness.
Generative AI Boom: Investment in tools like ChatGPT skyrockets ($25.2 billion in 2023), highlighting the technology's disruptive potential.
Regulation Accelerates: Growing public concern and AI's impact on jobs prompt a surge in US regulations, setting guidelines for responsible AI use.
๐ Read the report here
Stack Overflow and OpenAI Team Up to Supercharge AI Development
Stack Overflow, the world's largest developer community, is partnering with OpenAI (creators of ChatGPT) to enhance AI models.
Stack Overflow's vast knowledge base will improve OpenAI's accuracy and reliability, while OpenAI's tools will streamline development on Stack Overflow.
Why it matters: This collaboration signals a significant step towards better AI-powered tools for developers. Key benefits include:
Improved AI for Developers: OpenAI models will leverage Stack Overflow's trusted technical content, potentially leading to faster, more accurate solutions in tools like ChatGPT.
Enhanced Development on Stack Overflow: Stack Overflow will use OpenAI models to create better products for its users, driving community growth and engagement.
Community Recognition: OpenAI will attribute Stack Overflow contributions within ChatGPT, giving the community credit for their work.
๐ฐ Stack Overflow's press release
Director Uses SORA to Craft First AI Music Video, Offers Insights
Director Paul Trillo has created a music video for indie band Washed Out using OpenAI's AI video tool, SORA.
While the video showcases SORA's capabilities, it also highlights current challenges like high costs and the need for skilled prompt engineering.
Why it matters: This project offers a glimpse into the potential and limitations of AI video generation for creatives. Key takeaways include:
Impressive Results: Despite limitations, SORA generated cohesive, visually compelling footage for the four-minute video.
The Cost Factor: Even with recent optimizations, SORA remains potentially expensive, with estimates around $650 for the project. This could be a barrier for some creatives.
Prompt Power: Skillfully crafting text prompts is crucial for getting the desired results. SORA's lack of multimodal input (ability to use images as reference) adds to the challenge.
๐ค Interview with the director
Elon Musk's AI-Powered News Vision for ๐
Elon Musk plans to use AI to revolutionize how news is presented on his app, X. A new AI bot, Grok, will summarize news stories and incorporate real-time reactions from X users for a dynamic, comprehensive news experience.
Why it matters: This unique approach has the potential to disrupt how we consume news, but also raises concerns for publishers and presents legal challenges:
Social Media Focus: Grok's summaries will rely solely on X posts, making it distinct from other AI summarizers but potentially more controversial.
Citation Challenges: Accurate citations are crucial for ensuring user trust and providing a gateway to original sources. Publishers rely on these links for traffic and revenue.
Legal Minefield: Relying only on commentary could be a tactic to avoid copyright lawsuits that plague other AI news tools, but courts may not be sympathetic.
๐ฐ Article by Big Technology
OpenAI Tackles Content Authenticity with New Tools and Standards
OpenAI is taking steps to ensure the integrity of online content as the use of generative AI becomes more widespread. The company is joining the C2PA standards coalition, integrating metadata into its own AI-generated images, and developing new tools to detect and watermark AI-created content.
Why it matters: These efforts aim to help people understand the origins of online content and combat the spread of misinformation. Key advancements include:
Open Standards: OpenAI's support of the C2PA standard will make it easier to verify the source and authenticity of digital content, fostering trust online.
AI-Specific Tools: New detection classifiers help identify images generated by OpenAI's DALL-E 3 model (98% accuracy), while audio watermarking adds a layer of security to Voice Engine.
Collaborative Approach: The company recognizes the need for industry-wide collaboration to tackle challenges of content authenticity effectively.