Natural-sounding AI Chatbots, Assistants, and Robots!

Revealing the secrets behind the natural-sounding AI... from Neuralink to LLM building

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:

  • 📽️ Demoing the fastest Language User Interface (LUI) on CNN!

  • 🧠 Brain-controlled robotics using deep learning

  • 🖼️ AI sounds natural enough to be a therapist?

  • 📲 3 new trending AI Apps!

  • 🎤 New episode of the AI Minds podcast!

  • 👟 Building an LLM: Pre Training tricks and data processing tips

  • 🐺 Lone wolf versus the OSS community: Who produces better AI?

  • 🐦 Wired & Forbes argue AI-video has a brighter future than AI-chatbots

  • 📚 AI Glossary on Assistants, Mistral, and Backpropagation

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

Oh yeah, and while you may be familiar with Deepgram’s speech-to-text API, you might want to check out our upcoming text-to-speech technology as well 🥳

📽️ Showcasing a Language User Interface (LUI): Groq, Deepgram, & More on CNN

In this video, CEO of Groq Jonathan Ross demonstrates how speed can make AI interaction feel more natural. CNN’s Becky Anderson really puts the AI model through a rather drilling interview, and the technology prevails. 

PlayHT, Deepgram, Mistral, Vapi, and of course Groq are among the language model leaders making this natural-sounding AI as readily available as possible.

🧑‍🔬 Brain-controlled robotics and AI Therapists

Cognitive neuroscience and robotics: Advancements and future research directions - This article focuses on using AI algorithms, especially deep learning, to enhance Neuralink-like technology. Specifically, it looks at ways for humans to control robots with their minds via electroencephalography guided by deep learning and brainwave pattern recognition.

Depression Therapy Chat-Bot using Natural Language Processing - Are AI chatbots natural-sounding enough to be effective therapists? This paper believes so. In fact, it reports that “the chatbot therapy group experienced significantly fewer depressive symptoms than the control group (BDI-II). [And] high levels of satisfaction with the therapy and the chatbot interface were also expressed by participants in the chatbot therapy group.”

GetFloorPlan - Create 2D and 3D floor plans with AI! If you need high-quality visuals quickly, check them out!

Language Confidence provides API access to their leading-edge AI Speech Assessment Toolkit that automates the teaching and testing of spoken English for students. Your students can practice their spoken English anytime, anywhere.

Emma is an AI-powered virtual assistant that can be customized for your business in just minutes. With Emma, you can quickly build chatbots to automate tasks, provide customer support, and more.

🎙️ AI Minds Podcast!

Last week, we launched our AI Minds podcast! Featured below is Episode 2: Samiur Rahman, CEO & Founder at HeyDay. Check it out 😁

“I think anytime you try to automate away humans from something, it's going to seem really cool at first, but generally speaking, it's going to be like 80% good enough, and that's fine for a lot of things."

Samiur Rahman on automation in human jobs

🏇 Building an LLM Stack & Making it Feel Natural 

Building an LLM Stack Part 2: Pre-training Tips & Data Processing Tricks - Before you can build a chatbot, you have to start with an LLM. And in this article—part two of a series—Andy Wang walks us through the pretraining and data processing techniques needed to ensure your time building goes as smoothly as possible!

Lone Wolf vs Community: The Benefits of Open Source Software - One way to make your chatbot sound as natural as possible is to use open-source software. After all, the CNN demo above fully uses OSS tech, and if it’s good enough for national TV, it’s probably good enough for the rest of us!

🐝 Social media buzz

As an alternate perspective, Forbes argues that AI video generation is all the rage this year rather than AI Chatbots. Likewise, Wired is focusing on AI studying film rather than language & law. What do you think?

📚 Glossary Pages

The glossary entries below outline everything you need to know about each respective topic—from histories, to use cases, to current implementations & resources. Check them out!

AI Assistants - What’s the difference between reactive agents, deliberative agents, collaborative agents, and hybrid agents? What’s the difference between an agent and an assistant? Find out in this glossary entry!

Mistral - In the CNN interview above, Jonathan Ross cites Mistral as one of the OSS technologies Groq used to make the LUI sound as natural as possible. Learn more about Mistral here!

Backpropagation - Infamously one of the most mathematically difficult concepts to be introduced to when learning about AI, backpropagation has quite the reputation. This glossary entry simplifies it as much as possible, and even discusses how it's used today.

🖥️ Upcoming Webinar: An audio intelligence product showcase!

Tune in next week for the latest on Deepgram’s Audio Intelligence features and more product updates. Save your seat here!

Date: February 21st 2024

Time: 11am PT | 1pm CT | 2pm ET