ChatGPT Made Up His Career ChatGPT wrote Jonathan Schaeffer (an AI Researcher) an incredible biography. There was only one problem: 21 details were wrong. The AI-generated bio included impressive achievements and awards, and it sounded convincing enough that another person believed it. But Jonathan knew the truth because it was…
Would You Trust AI With 99% Accuracy? Jonathan Schaeffer uses this powerful analogy to explain why 99% AI accuracy may still be unacceptable. An artificial intelligence system can generate hundreds of correct answers, but one confident mistake could cause serious consequences. The biggest problem is that users often cannot identify…
AI Chatbots Won’t Last The AI chatbot may only be a temporary stage in the evolution of artificial intelligence. Rebecca Rosenberg predicts that people will quickly outgrow the current chatbot interface. Instead of typing prompts and interacting with AI as if they were emailing Claude, future AI systems may allow…
AI can sound confident, friendly, and brilliant, and still fabricate an answer. In this episode of An Hour of Innovation podcast, Vit Lyoshin speaks with Jonathan Schaeffer, founder and CEO of Synsira and creator of Kind, about AI trust, hallucinations, privacy, local AI, and why today’s most impressive tools should…
What happens when accessibility technology loses its internet connection? Rebecca Rosenberg (CEO and Founder of ReBokeh, a Personalized Vision Technology) explains why ReBokeh was built natively for both iOS and Android, despite the additional cost and development time. By optimizing the app for each platform, the team created a more…
Can Accessibility Features Can Make Apps Worse? Adding more accessibility features can sometimes make a product less accessible. Rebecca Rosenberg (CEO and Founder of ReBokeh, a Personalized Vision Technology) explains why tools like ReBokeh created for people who are completely blind do not automatically work for people with low vision.…
Accessibility Is Never One Size Fits All Personalized technology should give users control instead of assuming an algorithm always knows what they need. Rebecca Rosenberg explains why ReBokeh does not automatically assign accessibility settings based on a person’s diagnosis. Two users with the same visual condition can require completely different…
Artificial intelligence makes building new features easier, but that does not mean every feature should exist. Rebecca Rosenberg (CEO and Founder of ReBokeh, a Personalized Vision Technology) argues that one of the most important questions for AI founders and product teams is what they will intentionally refuse to build. With…
Does Vibe Coding Has a Security Problem? The traditional startup advice to ship quickly becomes riskier in the age of vibe coding. Rebecca Rosenberg (CEO and Founder of ReBokeh, a Personalized Vision Technology) explains why founders can no longer release AI-generated products without carefully considering usability, safety, data privacy, and…
A massive market can remain invisible when investors have never personally experienced the problem. Rebecca Rosenberg, CEO and founder of ReBokeh, a personalized vision technology company, explains how investors underestimated the low vision accessibility market even while speaking directly to a founder with low vision. Approximately 250 million people worldwide…
Why Voice Assistants Keep Failing We've been promised the perfect AI assistant for years. Siri, Alexa, AI pins, and other voice assistants were supposed to let us simply talk to computers and have them understand what we want. So why hasn't it worked? Carter Huffman (CTO and Co-Founder at Modulate,…
Most products do not fail because nobody needs them; they fail because teams misunderstand who they are building for. In this episode of An Hour of Innovation podcast, Vit Lyoshin speaks with Rebecca Rosenberg, founder of ReBokeh, about building visual-first accessibility tools for people with low vision and what her…
Sometimes an AI problem isn't actually an AI problem. It's hardware. Carter Huffman (CTO and Co-Founder at Modulate, a Voice Conversion technology company) explains how something as simple as an Ethernet cable created interference in analog audio signals used for machine learning data. Because the audio was converted from analog…
Why Voice AI Is So Easy to Break Voice AI can understand the words you say and still completely misunderstand the conversation. Carter Huffman (CTO, Co-Founder at Modulate, a Voice Conversion technology) explains why voice AI is one of the harder problems in artificial intelligence. Unlike text, human speech contains…
AI can understand you perfectly through one microphone and completely fail when the audio setup changes. Carter Huffman (CTO, Co-Founder at Modulate, a Voice Conversion technology) explains why Voice AI and machine learning models can be surprisingly sensitive to differences in microphones, audio quality, background noise, and recording environments. Humans…
How do you price a product correctly, and can AI determine what customers are actually willing to pay? Ravi Sawhney shares a surprising pricing strategy in which customers reportedly valued a product more at $150 than at $100. Understanding product valuation is essential for scaling. Learn how a simple price…
Robots Will Power the Space Economy The future space workforce may be mostly robots. Ethan Barajas (Co-Founder & CEO of Icarus Robotics) explains why robotic labor could become a massive part of the growing space economy. Instead of using highly trained astronauts for repetitive work, robots can handle menial tasks…
Why are human spaceflight costs so high? Cheaper rocket launches won't automatically make human spaceflight affordable. Ethan Barajas (Co-Founder & CEO of Icarus Robotics) explains why the true cost of sending astronauts to space goes far beyond the rocket. Mission control, safety requirements, testing, life support, training, and the infrastructure…
Can AI understand human emotion or can it only predict human behavior? Artificial intelligence can recognize patterns and anticipate your next action, but Ravi argues that it still cannot understand the human motivation, intention, and aspiration behind that action. Get practical product manager advice to help you build your next…
Robots Can Work Where Humans Literally Cannot Some places in space are simply too dangerous for humans to work. Ethan Barajas (Co-Founder & CEO of Icarus Robotics) explains how extreme radiation and other hazardous conditions create environments where astronauts cannot safely operate, but robots can. Space robotics and autonomous systems…
The Real Reason Advanced AI Models Still Fail Why do advanced AI agents still give wrong answers? Sid Bharath (AI Consultant and founder) explains that even the latest AI models cannot overcome bad data, missing context, or incomplete instructions. AI agents do not have human intuition. Whether you use ChatGPT,…
Master AI agent training through a practical show-and-tell approach. AI tutorials can explain what an AI agent is, but they rarely change how people work. Employees begin to understand AI automation when they see it complete their real tasks in seconds. Sid Bharath, AI Consultant and a Founder, explains why…
AI-Native Companies Still Need Humans Sid Bharath (AI Consultant and Founder) explains why becoming an AI-native company is about integrating artificial intelligence into business workflows—not handing every decision to AI. Companies still need human judgment to validate AI output, provide strategic direction, maintain quality, and decide which processes should be…
AI Agents Could Replace Business Apps What if AI agents make traditional business apps obsolete? Sid Bharath predicts that workers may eventually stop switching between CRM software, email marketing platforms, dashboards, and other business tools. Instead, one intelligent AI agent could access company data, connect with multiple systems, and complete…