in recent years, the world has witnessed tremendous development in the generative artificial intelligence field, and as we approach 2025, it has become clear that this field will continue its progress at an unprecedented pace. from radical changes in how companies use technology to new breakthroughs in space, here are 10 key predictions that will shape the landscape of artificial intelligence next year.
1. meta starts charging for the use of llama models
the shift from open source to a profitable business model
meta (the parent company of facebook and instagram) is one of the most prominent players in the field of artificial intelligence, distinguishing itself from competitors like openai and google by offering open-source llama models. this approach helped it achieve widespread fame, especially among independent researchers and developers.
however, with the ongoing financial challenges facing companies in developing large language models, it appears meta will shift its strategy. it is expected to start charging for the use of llama in commercial environments. this shift will not include all users, but will be limited to large organizations that use these models on a large scale to generate massive revenues.
what is driving meta to make this change?
- the exorbitant costs of developing advanced models: training advanced models like llama is a massive investment requiring billions of dollars annually. with the rising cost of cloud computing and data, continuing to provide these services for free becomes difficult.
- competition from major corporations: mark zuckerberg previously stated that giant companies like microsoft and amazon make substantial profits from reselling llama models. thus, charging a reasonable fee for the use of these models is considered a logical step to secure revenues.
- balancing open source and the profit model: despite charging fees, llama is expected to remain free for researchers, hobbyists, and small developers, maintaining its open-source character.
expected market impact
- this move will lead to radical changes in how artificial intelligence models are used, potentially prompting small and medium enterprises to look for open-source alternatives or develop their own models.
- on the other hand, it will help meta enhance its financial sustainability and expand its investments in developing future models.
2. new laws for expansion in robotics and biology
what are scaling laws?
retail expansion laws trace back to a 2020 study by openai, which indicated that increasing the number of parameters, data, and compute capacity significantly improves model performance. these laws have led to amazing developments in the field of language models like gpt-4 and gpt-5.
applying laws to new fields
until now, scaling laws have revolved around language processing. but 2025 will witness the application of these laws to other fields, most notably:
- robotics: developing artificial intelligence systems that enable robots to self-learn and adapt to new environments without intensive human intervention.
- biology: building artificial intelligence models that help discover new drugs and understand complex biological systems.
importance of these developments
- scaling laws will pave the way for breakthroughs in developing robots capable of performing precise tasks like surgery or underwater exploration.
- in biology, these laws will contribute to accelerating scientific discoveries and treating intractable diseases.
3. the breakdown of the relationship between donald trump and elon musk
impact on artificial intelligence policies
recent years have witnessed a rapprochement between former president donald trump and businessman elon musk. this rapprochement could significantly affect how the us government deals with artificial intelligence, particularly regarding regulatory policies and corporate support.
why might the relationship break down?
- trump and musk are complex personalities, each with different visions and strategies.
- history shows that trump often cuts ties with his allies shortly after a period of time.
repercussions
- the breakdown of the relationship could negatively affect companies like tesla, which rely on musk’s proximity to the us administration.
- on the other hand, openai might benefit from more neutral policies toward competition.
4. web agents as the next killer app
what are web agents?
web agents are artificial intelligence systems that can execute tasks across the internet on behalf of the user, such as paying bills, booking appointments, or managing subscriptions.
why now?
- advances in language models and computer vision have made these agents more accurate and efficient.
- these systems provide users with a practical solution that saves time and effort.
market impact
- web agents are expected to become one of the most successful applications in the field of consumer artificial intelligence.
5. space-based artificial intelligence data centers
the need for new solutions
with the rising energy demand to power data centers, the idea of establishing data centers in space has emerged. these centers will rely on solar energy and solve the cooling problem thanks to low temperatures in space.
challenges
- the high cost of launching equipment.
- the need for advanced technologies to transfer data between earth and space.
opportunities
- companies like lumen orbit have already begun exploring this idea, indicating promising potential.
6. the audio turing test
current challenges
despite the success of language models in passing the written turing test, significant challenges remain in simulating natural voice conversations.
what’s new in 2025?
- improvements in natural language processing and voice recognition technologies.
- reducing latency between the artificial intelligence response and the user.
7. artificial intelligence systems designing other artificial intelligence
expected progress
- research in this field will witness major progress, as models will be capable of self-improvement and creating more efficient models.
8. major companies shifting to application development
why?
- building large models is very expensive and commercially unsustainable.
- therefore, companies will focus on developing applications that generate direct revenues.
9. misinformation surrounding artificial intelligence use
klarna’s case
- klarna may face intense scrutiny over its exaggerated claims regarding its use of artificial intelligence, highlighting the importance of transparency in this field.
10. the first real artificial intelligence security incident
what is expected?
- we may witness a security incident consisting of an artificial intelligence model behaving unexpectedly, which will sound the alarm about the importance of artificial intelligence safety.
conclusion
as technology prepares to play a larger role in our daily lives, it is clear that 2025 will be a pivotal year in the evolution of artificial intelligence. these predictions will serve as major turning points, with significant challenges and opportunities alike.
