What lies ahead in 2025?

AI companions, humanoid robots, and beyond.

Welcome, AI enthusiasts

Happy New Year, and welcome to 2025! On Tuesday, we looked at how AI reshaped our world over the past year. Today, we’ll dive into what’s ahead — from autonomous AI agents and humanoid robots to groundbreaking scientific discoveries powered by LLMs.

Today’s Insights

  • AI agents and the rise of the arbitrage economy.

  • Surprising uses and AI companions

  • AI-savvy businesses, humanoid robots, and the path to AGI.

Read time: under 5 minutes

AI Forecasts for 2025

Satya Nadella is spearheading Microsoft’s efforts to develop AI agents capable of acting on behalf of users. Source: AP Photo.

AI agents are poised to be everywhere in 2025.

  • Tech giants like Microsoft, OpenAI, and Alphabet are advancing new models capable of understanding and interacting with what’s on your screen and beyond. These innovations will drive AI agents—intelligent software designed to handle tasks on your behalf seamlessly.

  • These agents will expand their roles into real-world applications, managing tasks such as placing orders, addressing customer support queries, and streamlining repetitive processes in workplaces.

  • What’s more, we’ll see multiple AI agents joining forces, leveraging their unique capabilities to tackle intricate and collaborative projects, reshaping productivity and problem-solving in unprecedented ways.

Agents are set to drive a new arbitrage economy

  • For certain tasks, such as data analysis and report generation, AI agents will surpass human capabilities.

  • Tech-forward companies are likely to favor AI agents over human workers for these roles. For instance, AI agents are already handling customer service calls at a fraction of the cost of employing real people.

  • Meanwhile, in less tech-savvy organizations—which make up the majority of the economy—resourceful employees will leverage AI agents to delegate tasks, enhance efficiency, and increase their productivity.

Exploring new ways to develop more advanced AI

  • By late 2024, startups noticed their latest models weren’t progressing as quickly as earlier iterations, prompting a shift toward innovative training methods.

  • Researchers are now focusing on techniques that use less data and are more energy-efficient. One such approach is chain-of-thought reasoning, which tackles complex problems by breaking them into smaller, manageable steps.

  • OpenAI’s reasoning-centric o3 model is leading this trend, inspiring others like Alphabet and Alibaba to follow suit.

  • We can also expect increased experimentation with SLMs (small, task-specific fine-tuned models), neuro-symbolic AI (which blends traditional machine learning with cutting-edge techniques), and other emerging methodologies.

Meta’s Orion glasses Source: Meta

AI will appear in unexpected places

  • In 2025, AI will become even more present as generative features are integrated into our phones and laptops. We may also see a rise in AI-powered devices like Meta's Orion AR glasses and smarter Alexa devices.

  • Doctors will increasingly rely on AI as a valuable assistant, not just for note-taking and scheduling but also as a second set of eyes for interpreting X-rays and scans, often with better accuracy than humans.

  • Scientists will turn to AI for advanced predictions, from long-term weather forecasting to discovering new treatments for diseases and identifying genetic issues that would otherwise take years to uncover.

We’ll develop deeper connections with our LLMs

  • Chatbot platforms like Character AI will grow even more captivating as AI becomes more adept at mimicking human behavior.

  • You'll be able to design your own AI companion, with personalized emotional and physical traits.

  • An increasing number of people will seek AI for therapy and mental health support (with over a third of Americans already open to the idea).

A new wave of voice apps will emerge

  • With AI now proficient in understanding and conversing in human language, voice interfaces will replace text for many tasks, offering faster and more intuitive interactions.

  • For example, instead of navigating through your phone to find a product to buy, you might simply speak your order, and the AI will take care of it for you.

1X’s NEO robot could soon help with household chores. Source: 1X.tech

Businesses will embrace AI, and demand for AI expertise will soar

  • In 2024, the use of generative AI among business leaders grew from 55% to 75%, but widespread adoption remains a challenge.

  • Moving forward, businesses will leverage their own data to create customized AI systems suited to their needs, enlisting both internal experts and external consultants to guide the transition.

  • AI startups will likely thrive, attracting more investments as the industry is projected to reach a value of $1.81 trillion by 2030.

  • AI literacy will become even more crucial: Educators are warning that students who don’t engage with AI may fall behind, and a recent IBM report suggests AI knowledge could soon be a requirement for many jobs across industries.

The rise of humanoid robots

  • The first AI-powered consumer robots will start hitting the market, designed to assist with tasks like folding laundry and washing dishes, though they’ll come at a high cost.

  • Manufacturers will increasingly rely on robots to automate physically demanding tasks, with humanoid robots from companies like Agility Robotics and Figure already working on factory floors for Amazon and BMW.

  • Realistic 3D simulations will allow robots to learn new skills quickly, enabling them to practice challenging tasks millions of times digitally before attempting them in the real world.

We’ll get closer to AGI

  • We’ll gain a deeper understanding of whether AI truly "thinks" independently and start peering into models’ decision-making processes.

  • OpenAI’s o3 model is already solving research-level math problems and performing coding tasks with the precision of a competition-level programmer.

  • Models will continue to improve at combining their various "senses," integrating linguistic, visual, and auditory skills to better understand the world around us.

  • The emerging consensus is that AGI may not come as a sudden breakthrough but through a series of incremental advancements. When it does arrive, it may not be immediately obvious, which is why companies like Microsoft and OpenAI are racing to define it before it becomes a reality.

Thank you for Reading.

Know someone who'd love this newsletter? Share the love.🌟

Reply

or to participate.