Welcome (back) to AI Minds, a newsletter about the brainy and sometimes zany world of AI, brought to you by the Deepgram editorial team.

In this edition:

• 🎥 Stanford’s Evo 2 is Generative AI for DNA Sequences

• 🚘 HERMES: Better Vision and Understanding for Self-Driving Cars

• 🤖 Why human-AI Relationships need Socioaffective Alignment

• 📚 How Karpathy Uses LLMs

• 🐦 Social Media Buzz: Should Open AI Be Worried about Manus?

• 💼 MIT on Analyzing Your Company’s “AI Maturity Level”

• ⚡ Technical Webinar: How to Build Responsive Voice Agents

• 🧑‍💻 Poised now turns your spoken words into working code!

• 📲 Three new, trending AI Apps for you!

• 🎙️ AI Minds Podcast with Simrat Singh, Co-Founder at Hooman Labs

Thanks for letting us crash your inbox; let’s party. 🎉

🎥 Stanford’s Evo 2: Generative AI for DNA

Imagine ChatGPT, but instead of writing words, it writes the DNA sequence required to produce a human with green eyes… or a person who’s likely going to grow over 6 feet tall… or someone with a natural talent for playing piano.

Do we have such technology already? Well perhaps it’s not too far out of reach.

“Stanford scientists have developed a generative AI tool that can predict the form and function of proteins coded in the DNA of all domains of life, identify molecules that could be useful for bioengineering and medicine, and allow labs to run dozens of other standard experiments with a virtual query – in minutes or hours instead of years (or millennia).

The open-source, all-access tool, known as Evo 2, was developed by a multi-institutional team co-led by Stanford’s Brian Hie, an assistant professor of chemical engineering and a faculty fellow in Stanford Data Science. Evo 2 was trained on a dataset that includes all known living species, including humans, plants, bacteria, amoebas, and even a few extinct species.”

🔍  How AI Cars Can Improve Their Vision and Why Human-AI Relationships Need Socioaffective Alignment

HERMES: A Unified Self-Driving World Model for Simultaneous 3D Scene Understanding and Generation - This paper presents a unified Driving World Model named HERMES, which integrates 3D scene understanding and future scene evolution (generation) through a unified framework in driving scenarios. It leverages a Bird's-Eye View (BEV) representation to consolidate multi-view spatial information while preserving geometric relationships and interactions.

Why human-AI relationships need socioaffective alignment - As AI capabilities advance, humanity faces a new challenge: the emergence of deeper, more persistent relationships between humans and AI systems. This paper explores how increasingly capable AI agents may generate the perception of deeper relationships with users, especially as AI becomes more personalised and agentic.

🐝 Social Media Buzz: How Karpathy Uses LLMs, and Should OpenAI Worry about Manus?

⚡ Technical Deep Dive: How to Build Responsive Voice Agents with Vonage & Deepgram

Learn how to build human-like voice agents for customer support, appointment scheduling and more in our March 26th technical webinar with Vonage.

When: Wednesday 26th March 2025, 10:00 PT / 12:00 ET / 17:00 GMT

Where: Online

Save your spot here!

Hosted by:

• Benjamin Aronov, Developer Advocate at Vonage

• Tony Chan, Senior Solutions Engineer at Vonage

• Damien Murphy, Applied Engineer at Deepgram

🔊 Poised for coding is here! Write code with your voice

Your days of hunching over a keyboard writing lines of code are officially optional. Poised now turns your spoken words into working code! There are two ways to code with your voice:

1. 🤖 Vibe coding: Speak naturally about what you want to build, hit a keystroke, and watch your words transform into code in Cursor, Replit, or your favorite AI editor. Lean back, sip your coffee, and create.

2. ⌨️ Interactive Coding: Speak naturally and watch your words convert to code. "Create a function that sorts an array of user objects by their signup date" → Boom. Done.

Plus, you can add new prompts or customize existing prompts to match your coding style and preferences. Say goodbye to carpal tunnel! 👍️ 

P.S. We'd love to feature your voice coding success stories. Share how you're using this feature by tagging us on social media (@poisedhq)!

📲 Three new, trending AI Apps for you!

Cleanvoice AI is an artificial intelligence software that removes filler sounds, stuttering, and mouth sounds from podcasts or audio recordings. It’s designed to save podcasters hours of tedious editing work.

Docalysis is an innovative AI-powered platform designed to transform how users interact with their documents. By leveraging artificial intelligence, Docalysis allows users to upload PDF, TXT, CSV, or DOCX files and engage in conversational queries to extract valuable insights and information without manually sifting through pages.

The Arduino Code Generator is a powerful tool designed for hobbyists and developers looking to create projects with Arduino boards. Leveraging the sophistication of OpenAI’s GPT-3.5-turbo language model, this generator provides a seamless way to create code tailored to specific Arduino tasks.

🎤 The AI Minds Podcast

Simrat Singh, Co-Founder at Hooman Labs, a startup building voice agents to help companies communicate with their customers over calls more efficiently. From consulting at McKinsey to private equity, Simrat realized he wanted a more hands-on role in building businesses, leading him to co-found Hooman Labs.

Simrat emphasizes that AI startups don’t need deep technical expertise but rather a clear understanding of how to solve real business problems using existing AI tools, prompt engineering, and API integrations.