Are AI Robots Creepy or Cool?

Does OpenAI's Figure 1 fall into the uncanny valley? Nobody will use AI if they find the bots creepy, researchers find.

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:

  • 🤖 Unintended consequences of humanoid service robots: A Case Study

  • 🏫 Cambridge conducts nationwide survey on whether AI robots are scary

  • 🏠 Are voice AI agents like Siri and Alexa getting creepier?

  • ⛰️ Does OpenAI’s latest robot (Figure 1) fall into the uncanny valley?

  • 🔊 Robot Voices: Top 11 Text-to-Speech models of 2024

  • 🐘 What is machine unlearning and why does it matter?

  • 🚗 NVIDIA, OpenAI, and Unitree Robot Demos

  • 📲 Three new trending AI Apps to check out!

  • 🎙️ NEW AI Minds Podcast with Fern Cowan!

  • 🤯 Bonus content: Comedian software devs react to OpenAI’s Figure 1 Robot

  • 👀 Bonus content: What is creepiness and what makes ChatGPT Creepy?

  • ⚒️ Tools that every AI Startup should be using

  • 📚 An in-depth glossary entry on Conversational AI

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

Deepgram just released a brand new text-to-speech model called Aura! Check it out here. 🥳

🧑‍🔬 Research: Nobody will use AI if robots feel creepy

Unintended consequences of humanoid service robots: A case study of public service organizations - Researchers at The University of Queensland conducted a study to better understand the unintended consequences of humanoid service robots. Their findings—taken from experiments conducted in a public library—theorize unintended consequences through dimensions of customer value and service interaction locus.

Scary Robots: Examining Public Responses to AI - In this comparatively old study, researchers from the University of Cambridge discuss the results of a nationally representative survey of the UK population on their perceptions of AI. Are people more optimistic or pessimistic about the future of tech? Find out here! (Note: 42% of respondents offered a plausible definition of AI, while 25% thought it [only] meant robots)

Perceived creepiness in response to smart home assistants: A multi-method study- We’ve discussed use cases of voice AI agents in a prior edition of our newsletter, but when it comes to familiar voice agents like Alexa and Siri, what do people really think will happen as home tech becomes more advanced? Will people continue to use these home devices if they know “too much”? And what does “too much” even mean?

📽️ Does OpenAI’s latest robot fall into the Uncanny Valley?

Watch Figure 1 not only showcase its impressive learned dexterity, but also its reasoning abilities. What are your thoughts on this video? Are you amazed or frightened?

How to make robots less uncanny

Top 11 Text-to-Speech models of 2024 - The more realistic your robot’s voice is, perhaps the more comfortingly human it becomes. At least that was our thesis in last week’s edition. And in that same vein, we present the top 11 text-to-speech models of 2024. After all, every robot needs a friendly voice!

What is Machine Unlearning and why does it matter? - Coined as "machine unlearning," this concept represents the converse of machine learning—it serves to make a model unlearn or forget. These algorithms, applied to previously trained models, force them to expunge specific portions of the training dataset. The beginnings of machine unlearning lie in responding to the "Right to be Forgotten" legislation, a provision of the European Union's General Data Protection Regulation (GDPR).

🐝 Social media buzz

NVIDIA GTC last week unveiled some incredible advancements, especially in robotics. Here are some social media snippets of these innovations!

AudioShake makes audio more interactive, immersive, and accessible by splitting recordings into their stems. Companies like OOONA, Cielo 24, Papercup, Singa, and Algoriddim have integrated AudioShake's technology to imagine new audio experiences in music, gaming, streaming, sports, UGC, and more.

Of course, physical bots aren’t the only bots out there. Landbot is a conversational experience platform that enables users to build chatbots with no coding required. The chatbots can be deployed on websites, WhatsApp, Facebook Messenger, and more.

Spot A Bot is a service that estimates the percentage of Twitter accounts that are bots for any Twitter trend. It analyzes the accounts tweeting about a trend and uses a machine learning model to determine which accounts exhibit bot-like behavior.

🎙️ AI Minds Podcast!

In this episode of AIMinds, we are joined by Fern Cowan, Founder at DeepCura, an AI medical scribe company. 

Fern shares his journey from culinary arts to videography, through San Francisco and Stanford, and ultimately into the healthcare industry. 

He discusses the inspiration behind DeepCura and the pain points it aims to solve, shedding light on the unique approach the company has taken. From the challenges of entering the healthcare industry to the intricacies of AI-assisted diagnosis and HIPAA compliance, 

Fern opens up about the hurdles and triumphs of building a product that physicians truly love.

🤖 Upcoming Webinar! Build AI Agents With Voice

AI agents that can understand and respond to users via natural voice communication are the future of customer interactions and personal assistance. In our upcoming webinar, Michelle Chan, Product Manager at Deepgram, will share her expertise on how to build efficient conversational AI workflows with voice.

Sign up here!

When: April 10, 2024 - 9:00am PT | 11:00am CT | 12:00pm ET

Where: Online

🤖 Additional Bits and Bytes

And, of course, if you’ve scrolled down this far, you get some (relatively random) juicy bonus content!

  • Comedians react to Figure 1 Robot - Strong language warning! Comedians and former software developers Cody Ko and Noel Miller react to the Figure 1 clip shown above. Be warned, this content is hilarious but rather crude. Viewer discretion is advised.

  • What is creepiness and what makes ChatGPT Creepy?  This article starts by reviewing studies into the psychology of creepiness, including a recent integrative account of the phenomenon. It proposes that the experience of being “creeped out” arises from threatening difficulties in understanding other minds.

  • Tools that every AI Startup should be using - This list goes over nine incredible AI Apps that startups everywhere should be using to improve productivity, increase growth, and maximize efficiency.

  • Glossary: Conversational AI- This article aims to unravel the intricacies of Conversational AI, from its core components to its evolving nature. How does this technology bridge the gap between human conversation and machine interpretation? Dive in to find out!