Our 206th episode with a summary and discussion of last week's big AI news!
Recorded on 04/07/2025
Try out the Astrocade demo <a href='https://www.astrocade.com/'>here</a>! https://www.astrocade.com/
Hosted by <a href='https://twitter.com/andrey_kurenkov'>Andrey Kurenkov</a> and <a href='https://twitter.com/jeremiecharris'>Jeremie Harris</a>. 
Feel free to email us your questions and feedback at <a href='mailto:contact@lastweekinai.com'>contact@lastweekinai.com </a>and/or <a href='mailto:hello@gladstone.ai'>hello@gladstone.ai</a>
Read out our text newsletter and comment on the podcast at <a href='https://lastweekin.ai/'>https://lastweekin.ai/</a>.
<a href='https://discord.gg/nTyezGSKwP'>Join our Discord here!</a> https://discord.gg/nTyezGSKwP
In this episode:
<ul>
<li>Meta releases LlAMA-4, a series of advanced large language models, sparking debate on performance and release timing, with models featuring up to 2 trillion parameters for different configurations and applications.</li>
<li>Amazon's AGI Lab debuts NOVA Act, an AI agent for web browser control, boasting competitive benchmarking against OpenAI's and Anthropic's best agents.</li>
<li>OpenAI's image generation capabilities and ongoing financing developments, notably a $40 billion funding round led by SoftBank, highlight significant advancements and strategic shifts in the tech giant’s operations.</li>
</ul>
Timestamps + Links:
<ul>
<li style="font-weight:400;">(00:00:00) Intro / Banter</li>
</ul>
Tools & Apps
<ul>
<li style="font-weight:400;">(00:01:46) <a href='https://techcrunch.com/2025/04/05/meta-releases-llama-4-a-new-crop-of-flagship-ai-models/'>Meta releases Llama 4, a new crop of flagship AI models</a></li>
<li style="font-weight:400;">(00:13:55) <a href='https://techcrunch.com/2025/03/31/amazon-unveils-nova-act-an-ai-agent-that-uses-a-web-browser/'>Amazon unveils Nova Act, an AI agent that can control a web browser</a></li>
<li style="font-weight:400;">(00:17:06) <a href='https://www.bloomberg.com/news/articles/2025-04-01/alibaba-preparing-for-flagship-ai-model-release-as-soon-as-april'>Alibaba Preparing for Flagship AI Model Release as Soon as April</a></li>
<li style="font-weight:400;">(00:17:59) <a href='https://techcrunch.com/2025/03/31/runway-releases-an-impressive-new-video-generating-ai-model/'>Runway releases an impressive new video-generating AI model</a></li>
<li style="font-weight:400;">(00:19:10) <a href='https://www.theverge.com/news/641270/adobe-premiere-pro-generative-extend-generally-available'>Adobe launches Premiere Pro’s generative AI video extender</a></li>
<li style="font-weight:400;">(00:20:54) <a href='https://www.testingcatalog.com/openai-prepares-reasoning-slider-and-memory-update-for-chatgpt-users/'>OpenAI prepares reasoning slider and memory update for ChatGPT users</a></li>
</ul>
Applications & Business
<ul>
<li style="font-weight:400;">(00:21:28) <a href='https://www.thehansindia.com/tech/nvidia-h20-chips-16-billion-orders-from-bytedance-alibaba-and-tencent-959631'>Nvidia H20 Chips: $16 Billion Orders from ByteDance, Alibaba, and Tencent</a></li>
<li style="font-weight:400;">(00:24:45) <a href='https://globalnews.ca/news/11104574/elon-musk-sells-x-xai-33-billion/'>Elon Musk sells X for $33 billion to his own AI startup company xAI</a></li>
<li style="font-weight:400;">(00:28:00) <a href='https://www.windowscentral.com/software-apps/softbank-dethroned-microsoft-as-openais-largest-investor'>SoftBank dethroned Microsoft as OpenAI's largest investor, pushing the ChatGPT maker's market cap to $300 billion — but reportedly buried itself in debt</a></li>
<li style="font-weight:400;">(00:30:48) <a href='https://arstechnica.com/ai/2025/04/deepmind-is-holding-back-release-of-ai-research-to-give-google-an-edge/'>DeepMind is holding back release of AI research to give Google an edge</a></li>
<li style="font-weight:400;">(00:34:06) <a href='https://wccftech.com/smic-5nm-development-completed-in-2025/'>SMIC Is Rumored To Complete 5nm Chip Development By 2025; Costs Could Be Up To 50 Percent Higher Than TSMC’s Version Due To The Use Of Older-Generation Equipment</a></li>
<li style="font-weight:400;">(00:36:04) <a href='https://www.pharmaceutical-technology.com/news/google-backed-isomorphic-labs-raises-600m-to-advance-ai-drug-discovery/'>Google-backed Isomorphic Labs raises $600m to advance AI drug discovery</a></li>
</ul>
Research & Advancements
<ul>
<li style="font-weight:400;">(00:38:03) <a href='https://arxiv.org/abs/2504.01848v1'>PaperBench: Evaluating AI's Ability to Replicate AI Research</a></li>
<li style="font-weight:400;">(00:43:50) <a href='https://arxiv.org/abs/2503.23829'>Crossing the Reward Bridge: Expanding RL with Verifiable Rewards Across Diverse Domains</a></li>
<li style="font-weight:400;">(00:48:39) <a href='https://arxiv.org/abs/2504.00294v1'>Inference-Time Scaling for Complex Tasks: Where We Stand and What Lies Ahead</a></li>
<li style="font-weight:400;">(00:54:34) <a href='https://arxiv.org/abs/2503.19206'>Overtrained Language Models Are Harder to Fine-Tune</a></li>
</ul>
Policy & Safety
<ul>
<li style="font-weight:400;">(00:58:28) <a href='https://deepmind.google/discover/blog/taking-a-responsible-path-to-agi/'>Taking a responsible path to AGI</a></li>
<li style="font-weight:400;">(01:02:32) <a href='https://www.nytimes.com/2025/04/03/technology/ai-futures-project-ai-2027.html'>This A.I. Forecast Predicts Storms Ahead</a></li>
<li style="font-weight:400;">(01:06:24) <a href='https://www.wsj.com/tech/ai/the-real-story-behind-sam-altman-firing-from-openai-efd51a5d?st=Qh5uBU&reflink=desktopwebshare_permalink'>The Secrets and Misdirection Behind Sam Altman’s Firing From OpenAI</a></li>
<li>OpenAI's new image generation capabilities represent significant advancements in AI tools, showcasing impressive benchmarks and multimodal functionalities.</li>
<li>OpenAI is finalizing a historic $40 billion funding round led by SoftBank, and Sam Altman shifts focus to technical direction while COO Brad Lightcap takes on more operational responsibilities.,</li>
<li>Anthropic unveils groundbreaking interpretability research, introducing cross-layer tracers and showcasing deep insights into model reasoning through applications on Claude 3.5.</li>
<li>New challenging benchmarks such as ARC AGI 2 and complex Sudoku variations aim to push the boundaries of reasoning and problem-solving capabilities in AI models.</li>
</ul>

Podurama

Home

Trending

Personal

Your Podcasts

New Episodes

My Files

Favourite Episodes

Tags (Folders)

Recently Played

Bookmarks

Followed Playlists

#206 - Llama 4, Nova Act, xAI buys X, PaperBench

Last Week in AI

AI Chapters

AI Insights

Reviews