Everyone, happy 2025! This is my first article of the year, and it's already off to a flying start—but I promise to resume publishing as frequently as usual. To kick off this new cycle, I want to share some of the hottest AI trends that MIT Technology Review has presented, which are set to shake up the market in the coming months.
1. Generative virtual worlds
After seeing AIs creating increasingly elaborate images and videos, the time has come to generative worldsImagine turning a simple sketch into a complete game or interactive 3D environment. Google DeepMind, with Genie 2, can already build entire universes from a single image—this opens up a huge range of possibilities, both for entertainment and for robotics research (training robots in virtual worlds).
Why is this relevant?
- New gaming experiences: Imagine creating an entire environment in seconds, testing mechanics and narratives.
- Collaborative: Scaled simulations help robots learn to navigate complex scenarios before going into the real world.
2. Language models that “reason”
The big change with models like OpenAI's new “o1” and “o3” is that they now “think” step by step instead of firing off the first answer that comes to mind. This approach reduces errors in math and logic calculations, for example, and is fundamental for the evolution of agents — those AI assistants that do research, take action, and are able to adapt when something goes wrong (like clicking the browser's “Back” button to check a detail, for example).
Why is this relevant?
- Less hallucinations: by breaking the problem down into steps, the model reduces the likelihood of unrealistic answers.
- More sophisticated applications: from programming to diagnostics, everything becomes more reliable.
3. AI advancing science
Scientific achievements via AI gained their greatest endorsement when the Nobel Prize in Chemistry was awarded to three researchers for projects involving the famous AlphaFold and AI tools. protein design. By 2025, we'll see even more research groups sharing data and training specialized models to solve problems in various areas, such as advanced materials e biology.
Why is this relevant?
- Accelerated discoveries: AI can test millions of scenarios and hypotheses in record time.
- New drugs and medical solutions: speeds up processes that previously took years, reducing costs and development time.
4. Partnerships (and dilemmas) between AI companies and national security
The military and security sectors are constantly investing in AI. Armed forces in several countries have been testing robots, drones, and data analysis tools for their operations. Startups and big tech companies are already eyeing this market, which is injecting billions of dollars into projects. By 2025, this search for partnerships is expected to grow up, which raises ethical debates about the use of AI in defense.
Why is this relevant?
- New sources of revenue for AI companies (and the resulting pressure to accept defense deals).
- Moral and political issues: push organizations to revisit their values and governance policies.
5. Nvidia under pressure from competitors
If, until recently, Nvidia was a choice when it came to AI hardware, this year other players like Amazon, AMD, and Broadcom have already shown their investments in powerful chips. There are also startups betting on alternative architectures that can surprise in terms of efficiency and cost.
Why is this relevant?
- Diversification: Greater competition creates more hardware options and can make access to technology less expensive.
- Geopolitics of chips: Tensions between the US, China and Taiwan influence export and production restrictions, fueling the quest for independence in semiconductor manufacturing.
And now?
For those who follow AI (or want to embark on this universe), 2025 arrives with unprecedented opportunitiesVirtual worlds will expand the boundaries of entertainment and research; "rational" agents and models will appear in virtually every software; and science will increasingly receive a boost from these intelligent tools.
Moreover, business and government decisions will increasingly demand accountability and transparency. Billions of dollars in profits are at stake, as well as ethical questions about how AI can (and should) be used.
Stay tuned, as this is only the first article of the year. I'll be publishing regularly again, always bringing perspectives, news, and insights on how AI is transforming the market—and our lives. What do you hope to see happening in the world of artificial intelligence in 2025? Leave your thoughts in the comments!
