Did OpenAI Steal Scarlett Johansson's Voice?

News, AI leaders, business insights and more

Hello AI Enthusiasts!

This Week’s Line-up

  • Did OpenAI Steal Scarlett Johansson's Voice?

  • Google Stunned the World at I/O Last Week

  • Meet Joelle Pineau, VP of AI Research at Meta

  • Rolls Royce: AI for Smarter Jet Engines

  • AI Chatbot, Assistant, or Agent: What's the Difference?

  • Learn, Connect and Grow

ps. Do connect with me on LinkedIn for daily news, insights and more. 😎

NEWS YOU CAN’T MISS

Did OpenAI Steal Scarlett Johansson's Voice?

Image: Paul Morigi/Getty Images

Scarlett Johansson, the actress who famously voiced an AI in the movie "Her," is now fighting against AI imitating her voice in real life. It's a scenario straight out of a sci-fi thriller, but this is no movie.

OpenAI, the company behind ChatGPT, approached Johansson to lend her voice to their new AI assistant, but she declined. Undeterred, OpenAI released Sky, an AI whose voice was so eerily similar to Johansson's that even her closest friends and family were fooled.

Even I, someone who follows AI developments closely, was convinced that OpenAI had secured Johansson's permission to use her voice. It was that good. But Johansson was "shocked, angered, and in disbelief" at the unauthorized use of her likeness, and her legal team quickly took action.

Why this matters:

This incident raises alarming questions about the ethical implications of AI technology and its potential for misuse. If even a celebrity like Johansson, with her resources and influence, can't protect her voice from being replicated and exploited, what hope is there for the rest of us?

OpenAI ultimately suspended the use of Sky's voice, but the damage was done. This controversy goes beyond a single actress and a single company.

It highlights the broader issue of how AI is rapidly evolving and the urgent need for society to establish clear boundaries and ethical guidelines to ensure its responsible use.

Google stunned the world at I/O last week.

Image: ProgrammerHumor.io

Google stunned the world at I/O last week with a jaw-dropping display of its latest AI innovations.

I've got the details below. But if you're short on time, the Salt Bae meme photo is basically Google's AI strategy in a nutshell.

My favorite highlights:

Search: When you "Google" something you are searching, you won't get a list of links to look up, you'll get summary answers of what you're looking for (this is pretty scary for those who rely on website traffic for business)

Meet Astra: Google's new AI assistant combines vision, understanding, and action to help with everyday tasks. Its voice has a personality, and you can interrupt it mid-sentence.

In a knockout demo, a woman walks around with her phone video on and then tells Astra that she can't find her glasses and it tells her that she left them on her desk next to the red apple. In other words, Astra has spatial awareness and can also "look" back at the video feed and answer your questions.

Gemini Everywhere: Google's flagship generative AI model will integrate with everything in its ecosystem, starting with Google Calendar, Tasks, and Keep (so imagine you want details from a party flyer added to your personal calendar - you can just tell Gemini to do that).

WOMEN IN AI

Meet Joelle Pineau, VP of AI research at Meta

Image: McGill

Meet Joelle Pineau, Vice President of AI Research at Meta and Professor at McGill University's School of Computer Science.

In 2019, the Governor General of Canada honored Pineau with the Innovation Award for her contributions to AI research. The Royal Society of Canada also elected her, recognizing her as one of Canada's most distinguished intellectuals and researchers.

Pineau excels in developing learning and decision-making methods for machines in uncertain environments. Her work in exploration and exploitation within reinforcement learning has earned recognition for its critical role in enhancing AI decision strategies.

Her contributions to using AI healthcare in particular have garnered praise: from personalized oncology treatment to robot-assisted health assistants, including the design of an intelligent robotic wheelchair (the Smartwheeler).

I absolutely support Pineau's call for researchers to open-source their code and focus on the importance of reproducibility in AI research. Reproducibility is key to the field's progress, enabling researchers to validate study results, ensuring their reliability and preventing cherry-picking. Pineau led the charge at NeurIPS, a leading AI conference, insisting that researchers include both code and comprehensive experiment details in their submissions.

News alert: Meta recently shifted Pineau's Fundamental AI Research (FAIR) lab to report to Chief Product Officer Chris Cox, moving from its previous reporting line to the CTO. This change aligns with Meta's vision to integrate advanced AI across all its products and services. Additionally, FAIR has narrowed its focus, discontinuing non-product-oriented research, including protein folding studies.

ENTERPRISE AI CASE STUDY

Rolls Royce: AI for Smarter Jet Engines

Industry: Aerospace

Image: Rolls Royce

Rolls-Royce isn't just building jet engines, they're building smart ones. In their quest to make aviation more efficient and reliable, they're harnessing the power of the Internet of Things (IoT) and AI.

A Network of Sensors in the Sky

Imagine a plethora of sensors embedded within each engine, constantly collecting data on everything from temperature and pressure to vibration and airflow. This massive influx of information is then sent to the cloud, creating a digital twin of each engine.

AI: Turning Data into Insights

With so much data at their fingertips, Rolls-Royce uses AI tools to analyze it all. This helps them predict maintenance needs, optimize flight paths for fuel efficiency, and even diagnose problems in real-time. By optimizing engine performance and reducing fuel consumption, Rolls-Royce is not only driving efficiency but also contributing to a more sustainable aviation industry.

PRO TIPS

AI Chatbot, Assistant, or Agent: What's the Difference?

Confused about the terms AI assistant, AI chatbot, and AI agent? Here's the breakdown:

AI Chatbot:

Purpose: Primarily designed for conversational interactions, often focused on answering questions or providing customer support.

Complexity: Can range from simple rule-based chatbots that follow pre-determined scripts to more sophisticated ones that use machine learning to understand user questions.

Interaction: With humans. Typically text-based, but can also incorporate voice.

Example: Those super annoying customer service bots on websites that are robotic, have no personality and know nothing beyond a very narrow scope of work.

AI Assistant:

Purpose: More sophisticated than chatbots with a broader range of tasks, which it can handle, like information retrieval, content generation, and controlling smart home devices.

Complexity: Utilizes AI to understand context, learn user preferences, generate content and provide personalized assistance.

Interaction: With humans. Can communicate via multimodal mediums (text-based, voice-based, image-based, video-based).

Example: OpenAI's ChatGPT, which blew everyone away because it could understand context and had an enormous library of knowledge that made conversations much more useful and smoother. Vintage models include the older iterations of Apple's Siri and Amazon's Alexa.

AI Agent:

Purpose: Built to perform a wider range of tasks autonomously, alone or in collaboration with other AI agents, including decision-making, problem-solving, and taking actions in the real world (e.g., booking flights, making purchases).

Complexity: Involves advanced AI techniques like reasoning, planning, learning, and interacting with complex and dynamic environments.

Interaction: Can interact with humans, other AI agents, and external systems.

Example: Recent demos from OpenAI's GPT-4o and Google's Astra have hinted at the tantalizing potential of AI agents.

🥇 And the winner for the AI agent du jour goes to … Astra from Google!

But hang on. Apple may have something up its sleeve on Jun 10 at its annual conference!

LEARN, CONNECT AND GROW

Week in AI: Stay Informed

What breakthroughs, controversies, and game-changers shook the AI world this week? I'll break it all down and reveal the implications for your business.

Join me on May 24 for my next Week in AI webinar.

AI Trendspotting: Long Read

From predicting player performance to keeping fans on the edge of their seats, the global AI in sports market is on track to hit $19.2 billion by 2030.

But how is AI making the experience better for everyone involved?

Read the full article and learn how Major League Baseball used AI to track player movements, how Formula 1 used AI simulations to optimize race strategies, and how AI helps athletes fine-tune their game.

See you next week!

-- The future awaits. Ayesha ♥️

I'd love your feedback...

Please vote below to help me improve the newsletter.

Login or Subscribe to participate in polls.