Welcome to Verses Over Variables, a newsletter exploring the world of artificial intelligence (AI) and its influence on our society, culture, and our perception of reality.

Tool Update

We talked about both of these tools last week, but the AI revolution marches on and ElevenLabs and Claude continue to improve.

ElevenLabs: Your Personal Audiobook Factory

What it is: ElevenLabs, the innovator in AI voice synthesis, is at it again. Their latest offering? The Reader App is a tool that transforms your dusty PDFs, websites, and endless newsletters into on-demand podcasts. Tired of generic AI voices? Enter Iconic Voices, where you can have Judy Garland, James Dean, or Burt Reynolds (yes, really) read you the news. It's like having a celebrity book club in your pocket, minus the cocktails and gossip. (The app is free for the first three months on iOs.)

How we use it: We've been force-feeding our morning newsletters into this digital narrator, turning our inbox into an audio playlist. Sure, our actual podcast queue is now longer than the line at Katz's Deli on a Saturday, but who cares? We're tackling our emails more efficiently, even if it means taking longer walks to finish listening. The things we do for productivity, right?

Claude: Your AI Butler Gets an Upgrade

What it is: Just when we thought Claude couldn't get any smarter, it goes and proves us wrong. Enter Projects, the organizational dream for every Type A personality out there. Think of it as an extended digital filing cabinet for your AI chats, but one that actually remembers where you put things, complete with customizable instructions and tone.

How we use it: We've been organizing our AI conversations with the enthusiasm of a Brooklynite color-coding their vinyl collection. We're essentially creating mini-GPTs as fast as possible. The result? A UI so organized it would make Marie Kondo weep with joy.

Some of our Favorite Tools

Krea: AI’s Real-Time Canvas

What it is: This real-time image generation platform is the digital equivalent of a hyperactive artist on a caffeine drip. Unlike its more patient AI cousins, Krea doesn't make you wait for your masterpiece — it evolves right before your eyes as you tweak your prompts. Krea's party trick? It lets you use images or drawings, doodles as prompts. Picture this: you scribble something that looks vaguely like a cat (no judgment, we've seen your drawings), and Krea transforms it into a majestic feline that would make Grumpy Cat look like an amateur. It's the artistic equivalent of turning water into wine, except it's turning your doodles into da Vincis. (Krea prompt window allows for text, your image reference on the left, and the live creation on the right.) Krea isn't content with just still images. It's branched out into video creation, turning your static ideas into motion pictures (with seamless keyframes). And for those of you still clinging to your low-res selfies, Krea's upscaling feature is here to drag you kicking and screaming into the world of high definition. Shockingly, Krea still has a free basic plan.

How we use it: We've been using Krea primarily for packaging and product design, because sometimes explaining your vision more challenging than herding cats. With Krea, we just show it what we want, and it's like it read our minds (note to self: check for hidden cameras). We've also dipped our toes into Krea's video creation tools. Fair warning: the render time is slow enough to make you question your life choices. But hey, Rome wasn't built in a day, and apparently, neither is your 10-second product video. Still, the results are worth the wait.

We’ll be talking about our favorite tools each week but here is a list of the tools we use most for productivity: ChatGPT 4o (custom GPTs), Midjourney (image creation), Perplexity (for research), Descript (for video, transcripts), Claude (for writing), Adobe (for design), Miro (whiteboarding insights), and Zoom (meeting transcripts, insights, and skip ahead in videos).

Intriguing Stories

AI: Artificial Infringement: In a growing legal conflict, AI companies like OpenAI and Microsoft face multiple lawsuits from media organizations. The New York Times and the Center for Investigative Reporting are among those accusing these tech companies of using their copyrighted content without permission or compensation to train AI models. The core issue in these lawsuits is that AI companies use vast amounts of internet content to train their models. This has raised significant copyright concerns, especially after Mustafa Suleyman, Microsoft's head of AI, controversially suggested that all content available on the internet is "free use" for training AI models. Complicating the matter, some media outlets have chosen to partner with AI companies, signing licensing deals with the AI companies. This has created a complex landscape where different media organizations are taking opposing stances on the issue. The controversy extends beyond just OpenAI and Microsoft. Perplexity, another AI search company, recently came under scrutiny for web scraping practices that appeared to violate website terms of service. As a result, Amazon Web Services (AWS) is now investigating Perplexity for potential violations of its terms of service. These developments highlight the ongoing tension between technological innovation in AI and the protection of intellectual property rights and raise important questions about the future of content creation and distribution in the digital age.

AI’s Economic Anti-Climax: As you can tell, we are fans of AI (or we wouldn’t be writing this newsletter), and Silicon Valley seems to be as well. But step outside that bubble, and you’ll find that the world is stubbornly resistant to the AI revolution. Despite big tech's $400 billion bet and investors' $2 trillion vote of confidence, AI's economic impact has been limited. When it comes to businesses actually incorporating AI into their processes, it's about as common as a flip phone at a Silicon Valley party. (We are fans of the new Motorola Razr and its AI-generated ad, but who’s taking notes.) The U.S. Census Bureau reports a mere 5% of businesses have used AI in the past month. Even in countries known for tech adoption like Britain, AI integration is progressing slowly. This hesitation stems from various concerns including data security, potential algorithmic bias, and a cautious approach to new technology. Many companies are experimenting with small AI projects, leading to what some call "pilotitis" - a situation where numerous small-scale trials make it difficult to identify the most effective areas for investment. Others are waiting for the technology to mature before making significant commitments. Those businesses implementing AI mainly focus on basic applications like customer service chatbots and personalized marketing. Even widely publicized examples of AI implementation, such as Klarna's staff reductions, may be more related to post-pandemic adjustments than revolutionary AI adoption. The expected productivity boost from AI is not yet visible in macroeconomic data. Most companies outside the tech sector are not significantly increasing their AI-related capital expenditure. However, it's worth noting that historically, transformative technologies often take time to be widely adopted and show economic impact. While AI's current economic effects are modest, it's too early to dismiss its potential for future impact. But for now, AI's economic revolution is less "War and Peace" and more "Waiting for Godot."

AI Gets a Hall Monitor: One of the significant challenges in AI development is the occurrence of hallucinations (instances where AI models generate answers that sound plausible, but are factually incorrect). (In one study of an older model, it hallucinated 21% of the time when asked factual questions.) To address this OpenAI has created a solution called CriticGPT. Think of it as the smart kid in class who's always catching the teacher's mistakes—except in this case, the "teacher" is ChatGPT, and the "mistakes" are bugs in computer code. CriticGPT, built on the bones of GPT-4, is designed to identify errors in code generated by ChatGPT. It's part of OpenAI's effort to improve AI alignment, ensuring AI systems behave as intended. The concept is straightforward: as AI assists in code writing, another AI system is needed to verify its work. In tests, human reviewers using CriticGPT performed better than those without it 60% of the time. This tool enhances the code review process, similar to an advanced spell-check for programming errors. Notably, CriticGPT identified mistakes in data that human reviewers had previously approved. However, CriticGPT isn't without limitations. OpenAI plans to incorporate it into its training process, aiming to improve the accuracy and reliability of future AI models. This represents progress towards AI systems that can potentially self-correct, albeit still requiring human oversight.

Byte-ing Off Too Much: Google's ambitious plan to go "net zero" by 2030 is hitting some serious turbulence. The tech giant's emissions soared by a whopping 48% compared to its 2019 baseline, with a 13% jump in 2023 alone. The increase is primarily attributed to the growing demand for AI and its energy-intensive data centers. As Google's Chief Sustainability Officer Kate Brandt puts it, they're navigating "a lot of uncertainty," especially regarding AI's environmental impact. The company's energy consumption has more than doubled in just four years, rivaling the output of power plants serving hundreds of thousands of households. Despite this, Google highlights that its data centers are 1.8 times more energy-efficient than the industry average. But experts aren't buying it. Lisa Sachs from Columbia University is calling for Google to step up its game, partner with cleaner companies and invest in a greener electrical grid. It's a classic case of "do as I say, not as I data." While Google maintains its commitment to sustainability, the company faces significant hurdles in balancing technological advancement with environmental responsibility. Google's green dreams are looking cloudy.

— Lauren Eve Cantor

thanks for reading!

if someone sent this to you or you haven’t done so yet, please sign up below so you never miss an issue.

if you’d like to chat further about opportunities or interest in AI, please feel free to reply.

if you have any feedback or want to engage with any of the topics discussed in Verses Over Variables, please feel free to reply to this email.

we hope to be flowing into your inbox once a week. stay tuned for more!

banner images created with Midjourney.