Your weekly briefing on all things AI

We know the rapid rise of AI is transforming our world, but with great power comes great responsibility. As OpenAI’s valuation skyrockets to $157 billion and new tools like “Canvas” reshape how we create, the line between real and artificial blurs further. Meanwhile, experiments like I-XRAY—smart glasses that can instantly reveal a stranger’s personal details—show just how easily cutting-edge tech can be weaponized. As AI becomes a dominant force in our daily lives, we must sharpen our skepticism and learn to differentiate what’s authentic and what’s augmented, or risk being fooled by fraudsters or even well intentioned hallucinations.

Let’s dive in:

> Invisible Threat: Smart Glasses Can Instantly Expose Your Identity

> How Eye Health Can Be a Window to Chronic Disease Detection

> OpenAI Raises $6.6B at a $157B Valuation

> Heatbot: Where User Experience Meets Intelligent Design

> Prompt Tip: Have Your Chatbot Evaluate Itself

Harvard Students Reveal the Dark Side of Smart Glasses

Two Harvard students have developed a demo showing how smart glasses, like Ray-Ban Meta, can use facial recognition technology to instantly reveal personal information like names, phone numbers, and addresses. Their project, called I-XRAY, streams video from the glasses to a program that scans faces and finds matching details from public databases. In the demo, they were able to identify classmates and strangers in real-time. The students say they don’t plan to release the tool but created it to raise awareness about privacy concerns. The demo highlights how easy it is to misuse smart glasses for surveillance, as their discreet design makes it hard to tell when someone is recording. It’s a reminder that personal information can still be accessed, even if you try to remove it from public databases. (Link)

More Stories

• Geoffrey Hinton, on the team of Tiferes portfolio company Cusp AI, was awarded a Nobel prize.  (Link)

• OpenAI Launches Canvas: A New Interactive Workspace for ChatGPT. (Link)

• Nvidia unveils a ChatGPT competitor that rivals GPT-4's capabilities. (Link)

• Meta is set to enter the AI video space with Movie Gen launching on Instagram in 2025. (Link)

• Microsoft launches Drasi, poised to revolutionize big data management. (Link)

• Anthropic welcomes OpenAI co-founder Durk Kingma to the team. (Link)

• OpenAI’s co-lead on Sora departs for Google’s AI division. (Link)

• Google Lens now answers video-related questions. (Link)

• Microsoft Copilot adds screen reading, deep thinking, and voice capabilities. (Link)

• Google Cloud integrates Search and YouTube technology into enterprise AI applications. (Link)

• Vera AI’s ‘AI Gateway’ helps businesses safely scale AI with less risks. (Link)

• AI-powered Raspberry Pi camera module enhances vision-based applications. (Link)

• Squarespace refines its AI tools with an emphasis on curation and taste. (Link)

• Math genius dubbed the 'Mozart of Math' is unphased by AI's rise in the field. (Link)

• AI chipmaker Cerebras prepares for IPO to challenge Nvidia in the chip market. (Link)

|| Healthcare

How Can Your Eyes Help Detect Diseases Before They Develop?!

A groundbreaking position paper in the Asia-Pacific Journal of Ophthalmology reveals how AI is poised to transform eye care and enhance cardiovascular health. By analyzing fundus images of the retina, AI can detect elevated HbA1c levels, a crucial marker for diabetes and cardiovascular risk, traditionally identified through blood tests. Early findings suggest that AI model performance can be affected by sample biases, emphasizing the need for reliable systems. Researchers stress the importance of responsible AI practices to improve patient care and promote innovation at the intersection of engineering and healthcare. This collaboration could transform chronic disease management and preventive healthcare. (Link)

|| Real Estate

Ryan Serhant: AI's Role in Shaping Empathetic Real Estate Agents

Artificial intelligence is transforming the real estate industry by automating tasks like property listings and market analyses, freeing agents to focus on personalized client interactions. Ryan Serhant, CEO of Serhant, explains that AI is forcing agents to build stronger relationships and find new ways to stand out in an increasingly competitive market. His company’s AI service, “Simple,” automates over 60% of sales tasks, allowing agents to focus more on personalized client attention. While AI streamlines lead generation and marketing, it complements the agent's role rather than replacing it. As AI gains traction, the future of real estate will see empowered agents delivering more tailored services, benefiting buyers and sellers alike. (Link)

|| Finance

Forbes Says Emotional AI Set To Disrupt Finance Sector

The emergence of Emotional AI in finance signals a paradigm shift in customer engagement and service delivery. The integration of emotional analytics could allow banks to tailor their offerings based on real-time assessments of a customer's emotional state. This evolution marks a new frontier in personalization, enabling financial institutions to provide more empathetic support and targeted advice during volatile market conditions. While the technology promises deeper customer loyalty and improved satisfaction, it also raises significant ethical and privacy concerns, highlighting the need for transparent practices and regulatory guidance. As banks explore this new terrain, balancing innovation with trust will be crucial in defining the future of emotionally intelligent banking. (Link)

• Equinix raises $15B to expand xScale data centers, addressing the growing demand for AI. (Link)

• OpenAI secures $6.6B of VC funding at a $157B valuation, solidifying its position in AI innovation. (Link)

• Poolside secures $500M in funding from eBay, Nvidia, and others for AI coding advancements. (Link)

• Crescendo’s AI contact center startup reaches a $500M valuation, rivaling traditional centers. (Link)

• Cancer AI Alliance launches with $40M to merge medical and tech expertise for next-gen care. (Link)

• Qodo raises $40M in Series A to elevate enterprise-grade code generation and testing. (Link)

• Resolve AI, backed by Greylock, secures $35M to enhance engineering workflows. (Link)

• Numa raises $32M to introduce AI automation to car dealerships. (Link)

• Series, a GenAI game platform, quietly raises $28M from Netflix, Dell, & a16z. (Link)

• Artisan raises $11.5M to deploy AI-powered ‘employees’ for sales teams. (Link)

• Avarra secures $8M to revolutionize avatar-based sales training with AI. (Link)

• Indian mental health app Lissun embraces AI with $3M in funding. (Link)

Introducing Heatbot: Where Intelligent Design Meets User Needs

Heatbot.io acts as your personal assistant for website design, harnessing AI to enhance user experience. By analyzing user behavior through heatmap data, it automatically generates optimized designs that resonate with your visitors. This tool is a game-changer for web designers, developers, and business owners, simplifying the process of boosting engagement and conversion rates. With a focus on personalization and accessibility, Heatbot helps you make informed decisions that genuinely elevate your website's performance, ensuring a seamless and engaging experience for every visitor.

More tools to try:

• Podial: transform documents into captivating podcast chats for effortless learning of complex topics.

• Graphy: master data storytelling by simplifying data presentation and communication.

• Jogg AI: convert URLs and product assets into engaging videos with diverse AI avatars and rich templates, driving website traffic and sales seamlessly.

• Replio: transform surveys into engaging AI-powered interviews.

• Robot Lawyer: revolutionize legal services with fast, affordable, and reliable AI-generated documents—verified for your peace of mind.

• Buzzabout: uncover key insights from online discussions, revealing what consumers truly think and feel about your product.

• RedactAI: generate high-quality LinkedIn posts that perfectly match your style with our AI tool.

• CustomerFinderBot: discover new customers with this AI bot that scans Twitter and Reddit for ready-to-buy leads—LinkedIn support coming soon!

• SocialSonic: unleash viral content and engage influencers on LinkedIn—all with one powerful AI platform.

• Character SDK: develop AI characters that can listen, speak, see, and act in real time.

• Singulatron: unlock rapid, intelligent decisions by connecting your team to a secure in-house AI hub for unparalleled collaboration.

• Buildpad: launch a minimum viable product with AI step-by-step guidance.

Have Your Chatbot Evaluate Itself

One useful way to ensure accurate outputs (although not full-proof) is to have ChatGPT review its own work. After generating a response, prompt it to perform a quick fact-check or research session to confirm the validity of its previous answer. For paying users, the phrase “do some research” often works to trigger web searching. Using a prompt like “do some research on [topic of x] and evaluate your last output for [criteria x]” can save time and improve reliability.

Pro Tip: Take it a step further by creating a customized evaluation framework. Develop a list of criteria for quality, tone, or content that ChatGPT can save in its memory. (Memory features need to be turned on). With a specific command word, you can then streamline these evaluations, making your chatbot even more aligned with your goals.

Example prompt: “Whenever I type $crossref$ before or after a url can you please do some research on the url and provide other sources for the same story?”