If You’re Not Using AI for These 15 Tasks, You’re Falling Behind (And Here’s Where to Be Careful)

Good morning. It’s Vlad here!

It’s Wednesday, December 11th, and today’s edition of the ImpactingAI Newsletter is packed with insights to help you stay ahead in the AI revolution.

Here’s what’s on the docket:

• 15 Times to Use AI—And 5 Times You Definitely Shouldn’t
• Are Multi-Agent Systems in AI Worth the Hype?
• 12 Days of OpenAI: Day 2, OpenAI Introduces Reinforcement Fine-Tuning for o1 Models
• And much more!

Let’s dive in!

Ethan Mollick’s article offers practical advice on where AI shines—and where it falls flat. If you’ve ever wondered whether to trust AI with a task, this is the guide you need.

When AI Can Help:

• Stuck on ideas? AI can supercharge your brainstorming.
• Too much to read? Let AI condense it for you—but don’t skip the critical details.
• Need fresh perspectives? AI can reframe content for different audiences or generate alternative phrasings when you hit writer’s block.

But Beware:
There are moments when AI can’t deliver. For instance, high-stakes decisions or ethical calls require something only humans can provide.

Want to know the full list of when (and when not) to use AI? Click below!

Multi-agent systems promise powerful collaboration between AI agents—but are they delivering on that promise? The article “Are Multi-Agents Really Useful?” explores whether these systems live up to their potential or fall short in practical use.

Key Takeaways:

• Complexity vs. Efficiency: Managing multiple agents introduces coordination challenges that can hinder performance.
• Real-World Impact: Despite their theoretical advantages, multi-agent systems often fail to outperform single-agent models in practical applications.
• High Costs: These systems demand significant computational power and development time, raising questions about their feasibility.

The Bottom Line:
Multi-agent systems offer intriguing possibilities, but their real-world utility remains uncertain. Is the added complexity worth the investment?

OpenAI has unveiled Reinforcement Fine-Tuning (RFT) for its o1 models—a breakthrough in AI customization. This new approach lets users fine-tune models to tackle specialized tasks with unparalleled reasoning capabilities.

What’s Different About RFT?
RFT doesn’t just teach models to mimic inputs; it trains them to think smarter by using feedback to refine their problem-solving skills. With just a few examples, models learn to generalize strategies, making this technique faster and more efficient than ever.

Why It Matters:
RFT is a game-changer for industries like:

• Legal: Automating complex workflows with precision.
• Healthcare: Analyzing rare genetic diseases with remarkable accuracy.
• Finance and Engineering: Driving better decision-making in data-heavy fields.

Curious how RFT works or what it can achieve?

Sora is here

OpenAI has launched Sora, a text-to-video AI model that generates realistic videos from text prompts. Available to ChatGPT Plus and Pro subscribers, Sora offers features like video generation up to 1080p resolution and 20-second durations, with options to extend, remix, and blend content.

Goodbye, unreliable weather forecasts? Google DeepMind's AI model sets new benchmark for 15-day predictions

Google DeepMind has introduced GenCast, an AI-driven weather forecasting model that delivers highly accurate 15-day forecasts. Trained on four decades of historical weather data, GenCast outperforms traditional models like the European Centre for Medium-Range Weather Forecasts' ENS, achieving over 97% accuracy in various scenarios.

Microsoft's AI CEO On How the Technology Will Impact Culture

Mustafa Suleyman, CEO of Microsoft AI, envisions AI companions as persistent, emotionally supportive entities that enhance user engagement. He emphasizes the importance of trust, security, and privacy in developing AI that can perform complex tasks autonomously, aiming to transform human-AI interactions across various platforms.

Apple hints at radically different AGI view than AI competitors

Apple's AI chief, John Giannandrea, has expressed skepticism toward the pursuit of artificial general intelligence (AGI), describing efforts to scale current technologies to achieve AGI as "very naive." Instead, Apple focuses on integrating AI advancements to enhance user-facing products, prioritizing practical applications over speculative goals.

AI Tools To Try