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.

The AI Hype Cycle: The Generalist’s Advantage

Connecting the Dots: Why Generalists are Thriving with AI

Dan Shipper, the CEO and co-founder of Every, is one of our favorites when it comes to AI. His weekly column, "Chain of Thought," and podcast "AI&I" have become must-reads (and must-listens) for us. Shipper isn't just pontificating from the sidelines—he's also in the trenches, experimenting with AI tools and pushing the boundaries of what's possible. In his latest piece, Shipper spotlights an unexpected victor in the AI revolution: the generalist. It's a topic that hits close to home for this intellectual nomad. From philosophy major to tech entrepreneur, from writer to podcast host, Shipper embodies the modern Renaissance person. And now, he argues, it's the generalists' time to shine.

Shipper contends that generalists are more than just people with shallow expertise in multiple areas. Their true strength lies in navigating uncertainty, connecting disparate ideas, and solving novel problems. This skillset, he suggests, is particularly valuable in "wicked" environments – complex situations where rules are unclear and feedback is often delayed or inaccurate. While AI excels in specialized, well-defined domains, it struggles with the creative problem-solving that generalists excel at. Shipper posits that AI tools like large language models empower generalists, allowing them to acquire new knowledge and apply insights from various fields quickly. Drawing an intriguing parallel to ancient Athens, where citizens were expected to wear multiple hats in civic life, Shipper speculates that AI might enable a return to this ideal of the multifaceted individual. Ultimately, Shipper's article paints a future where success isn't determined by having all the answers, but by knowing which questions to ask.

Full disclosure: we might be a tad biased here. As proud generalists ourselves, we're practically giving Shipper a standing ovation. But in a world where AI is eyeing specialist jobs like a hungry predator, perhaps our human edge truly lies in our ability to intellectual hopscotch. So go ahead, wave your generalist flag high. In the age of AI, knowing a little about a lot might just be the ultimate superpower.

Lessons from the Digital Frontier

You might dismiss our AI ramblings as the fevered dreams of a tech-obsessed futurist, but when Atlassian—the folks who keep half of Silicon Valley organized—embarks on a digital odyssey, it's time to sit up and take notice. The collaborative software virtuoso recently plunged 25 of its brightest minds into an AI immersion program, emerging with a playbook that's part user manual, part philosophical treatise.

At the heart of their discoveries lies AI's peculiar personality: imagine a savant with a photographic memory but the social graces of a caffeinated teenager. Atlassian advocates for a nuanced approach—leveraging AI's strengths while acknowledging its limitations.

Here are five key takeaways from their digital expedition:

1. Know Your AI's Quirks: Embrace AI's data-crunching prowess, but don't expect it to write the next Great American Novel.

2. Precision is Key: Vague requests yield vague results. Think less "write me a story" and more "craft a haiku about AI in iambic pentameter."

3. Iterate, Don't Hesitate: Start small, fail fast, learn quickly. It's less about perfection and more about progression.

4. ROI is King: If your AI assistant isn't boosting efficiency, it might be time for a digital performance review.

5. Context is Queen: The more background you provide, the better AI performs. It's not a mind reader (yet).

Atlassian's AI odyssey isn't about creating digital overlords, but crafting digital collaborators. It's a reminder that we're all still learning the steps in the dance between human and machine intelligence. So, whether you're an AI enthusiast or a skeptic, take a page from Atlassian's book—approach AI with curiosity, caution, and a willingness to adapt.

Tool Update

AI Takes on Data Journalism: Claude vs The Pudding

The Pudding, known for its innovative data-driven stories and our favorite site for data visualizations, recently embarked on an intriguing experiment: could an AI create one of their signature visual essays? They turned to Anthropic’s Claude and its Artifacts to find out.

The team guided Claude through their typical story creation process, from idea generation to final development. They aimed to replicate how a non-expert might interact with the AI, providing minimal technical assistance. Claude showed promise in several areas. It quickly generated story ideas, wrote functional code for data collection, and produced a solid storyboard. The AI excelled at isolated tasks, often completing them faster than a human might. However, the experiment revealed AI's current limitations. Claude struggled with complex problem-solving, often making mistakes that required human intervention to correct. It lacked the critical thinking skills to question results or identify edge cases, steps crucial to The Pudding's rigorous approach. The final product, while coherent, fell short of The Pudding's usual standards. The team concluded that AI tools like Claude can be valuable assistants, but they're not yet ready to replace human expertise in data journalism. This experiment offers a glimpse into AI's potential in creative fields. While it can handle many tasks efficiently, the nuanced decision-making and iterative refinement that elevate good stories to great ones remain firmly in human hands – for now. (If you’d like to read Claude’s attempt at data journalism, the story is linked to The Pudding’s report card.)

We’ll be talking about our favorite tools, 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

Apple’s Glow Up: More Tease than Dazzle: Apple's "It's Glow Time" event on Monday was less fireworks and more sparklers, leaving tech enthusiasts with a familiar itch to upgrade and a side of FOMO. The Cupertino giant unveiled its latest lineup of iPhones, Watches, and AirPods, but the much-anticipated "Apple Intelligence" felt more like a whispered promise than a showstopper. While the iPhone 16's camera upgrades stole the spotlight – think Hollywood-worthy spatial video capture and AI-powered photo editing that could make even your awkward selfies Instagram-worthy – the real star, Siri's AI makeover, seemed to be fashionably late to its own party. Apple's "coming soon" tease for advanced AI features left us feeling like kids peeking at wrapped presents before Christmas. Yet, in true Apple fashion, they've managed to create an irresistible allure. The beefed-up Siri hints at a future where your phone isn't just smart, but downright prescient. Imagine a digital assistant that can draft emails in your voice, summarize your day, or even help you win arguments with obscure facts (use responsibly, folks). And let's be honest – we'll all be lining up, credit cards in hand, ready to be part of whatever comes next. Because in the world of tech, FOMO is the ultimate upgrade motivator.

From Chatroom to Boardroom: AI chatbots are having their moment in the spotlight, and it's not just a fleeting cameo. These digital dynamos are quickly becoming the MVPs of the corporate world, amassing user numbers that would make even TikTok influencers blush. Meta AI is leading the charge with an impressive 400 million monthly active users. That's a lot of conversations happening across Facebook, Instagram, and WhatsApp – proving that AI isn't just a buzzword, it's becoming as commonplace as your morning scroll through social media. Not to be outdone, ChatGPT is holding its own with over 200 million monthly active users. But here's where it gets interesting: ChatGPT has made significant inroads into the business world, boasting over a million paying business users. That's a leap from 600,000 in April to over a million in just a few months – growth that would make any startup founder weak at the knees. From Arizona State University to Morgan Stanley, organizations across various sectors are embracing ChatGPT's Enterprise, Team, and Edu versions, as 92% of users report increased productivity, 88% are saving time, and 75% are seeing improvements in creativity and innovation. It's like having a super-efficient, endlessly creative colleague who never needs a coffee break. Now, if you happened to catch a recent Goldman Sachs chart showing a drop in monthly visits to ChatGPT's website, don't panic. It turns out the researcher made a rookie mistake: they forgot to update the URL when OpenAI changed from chat.openai.com to chatgpt.com. Oops. In reality, usage has skyrocketed to 2.6 billion visits in August. It's a reminder that even in the world of high finance, sometimes you need to double-check your homework.

AI Sticker Shock: OpenAI, the company behind ChatGPT, is orchestrating a potential pricing overhaul that could hit notes as high as $2,000 per month for its next-generation AI models, "Strawberry" and "Orion." But before you start pawning your vinyl collection, consider this: we might be witnessing a classic case of expectations management, Silicon Valley style. This eye-watering figure—a whopping 100x jump from the current $20 ChatGPT Plus subscription—could be OpenAI's way of tuning our price expectations to a higher frequency. By floating such a stratospheric number, the company might be setting the stage for a more palatable, yet still premium, price point. Imagine the collective sigh of relief (and the flurry of credit card swipes) if the actual price lands at, say, $500 per month. Suddenly, it feels like we're getting a bargain on bleeding-edge AI. The new models promise enhanced reasoning abilities, with Strawberry designed to tackle tasks that would leave current AIs scratching their digital heads. Orion remains shrouded in mystery, but expectations are crescendoing. OpenAI's strategy comes at a pivotal moment. Despite facing financial pressures, the firm is attracting attention from tech giants like Microsoft, NVIDIA, and Apple, potentially boosting its valuation to over $100 billion. Whether this is clever marketing or a genuine reflection of AI's evolving value, OpenAI is clearly rewriting the script for AI pricing.

AI’s Self-Aware Sensation: The launch of the new open source model dubbed Reflection 70B just gave us a masterclass in digital drama, with a side of silicon snake oil. Reflection 70B was announced as your average model—it's an AI with a mirror and a penchant for self-improvement. The brainchild of HyperWrite, the model supposedly uses a nifty trick called Reflection-Tuning. A "self-reflection” method is used to reduce hallucinations or wrong answers — it effectively stops to “think” before giving its final answer. Picture an AI that ponders its outputs like a chess grandmaster in Washington Square Park, second-guessing every move. This model promises to catch its flubs faster than a New Yorker spots a tourist in Times Square, boasting a mind-bending 99.2% accuracy on AI benchmarks. But even digital divas have their off days. When Reflection 70B first hit the open source scene, it stumbled harder than a newcomer on subway stairs. Cue the Twitter storm and Reddit roast. Yes, AI bros decided to eat their own. Faster than you can say "tech pivot," Matt Shumer and his HyperWrite team claimed to pull off a coding all-nighter worthy of a startup legend, allegedly fixing bugs and tweaking bugs faster than you can say “Pied Piper,” and Reflection 70B was back to outshining AI heavyweights. But whispers in the digital wind suggest this might be more smoke and mirrors than genuine AI sorcery. Is Reflection 70B a groundbreaking leap or just exceptionally good at acing its own pop quiz? (Word is still out if this is a grift or actually a breakthrough.) As the tech world holds its collective breath, Reflection 70B stands as a Rorschach test for the AI community. Is it a testament to rapid innovation or a cautionary tale of digital snake oil? In the high-stakes poker game of artificial intelligence, sometimes it's hard to tell if you're witnessing a royal flush or just a really good bluff.

AI’s Craft a Virtual Civilization: Imagine unleashing a cohort of hyper-intelligent AI agents into a Minecraft world and watching them build a society from the ground up. That's precisely what Altera did with Project Sid — a fascinating blend of SimCity and digital anthropology that's pushing the boundaries of artificial intelligence. These aren't your average Non-Player Characters mindlessly chopping wood or mining ore. Project Sid's AI agents are equipped with advanced social skills: they engage in nuanced conversations, form complex relationships, and even ponder abstract concepts. It's as if the cast of "Westworld" decided to colonize a digital frontier. Left to their own devices, these silicon settlers didn't just eke out an existence—they flourished. They spontaneously birthed an economy, with gems serving as the cryptocurrency of choice in their new marketplace. Democracy blossomed in the pixelated soil, complete with a constitution and voting system. One can almost picture diminutive AI lobbyists schmoozing in smoke-filled rooms constructed entirely of Minecraft blocks. The Project Sid team faces some unique challenges. How do you measure the success of an artificial society without inadvertently playing digital god? And how do you balance the agents' individual goals with the need for social cohesion? It's like trying to herd cats, if the cats were sentient and prone to forming their own system of government. As Project Sid continues to evolve, it's providing valuable insights not just into artificial intelligence, but into the nature of society itself. By observing how these digital citizens organize themselves, we might gain fresh perspectives on our own social structures and behaviors, and whether or not to pull the plug.

— 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.