Predictions for 2026
Let us play oracle and try to predict what the next year will bring in the world of technology.
The year 2025 is coming to an end. Like everyone else, I am thinking about what the new one will bring. Will it be just as chaotic and uncertain, or will things calm down a little? The only thing I am sure of is that it will be very different from what we have experienced so far. Everything else is speculation that I will likely laugh at in a year. I will focus only on technology and leave stuff like politics and economics aside.
Let's go.
AGI Is Not Coming
The first topic to debate is artificial intelligence. In recent years, we have seen a massive leap in what computers can do. This has surprised even strong skeptics. David Deutsch, the intellectual father of quantum computing, admitted that he was mistaken when he believed that a program capable of human-level language use must necessarily be AGI, meaning artificial general intelligence comparable to humans. Large language models (LLM) such as ChatGPT proved him wrong. The core mistake was the assumption that human language is inherently extremely complex and requires cognitive abilities on the level of the human brain. That assumption turned out to be false.
This unexpected partial success injected false optimism into many people. They assumed that general intelligence can be achieved in the same way. LLMs are language models. They have been surprisingly successful at what they were designed for, which is language processing. That does not mean they are suitable for other domains. Examples include driving a car, playing chess, or solving complex problems that require a holistic approach and deep contextual understanding. Language processing is still a specialization.
The year 2025 has already shown that scaling LLMs has limits. Improvements in their capabilities have largely stalled; at least we don’t see any progress by orders of magnitude anymore. This means that LLMs are not a path to AGI. AGI will require a completely new approach and genuinely novel ideas. Futuristic visions such as those promoted by Ray Kurzweil therefore look premature. Although companies like Meta and Google are also experimenting with non-LLM approaches to AGI, so far, there are no known groundbreaking results. I strongly doubt that these organizations would resist informing their stakeholders if they had a working AGI in the lab.
AI Products Will Emerge
But don’t get me wrong: The absence of AGI is not a failure. It may even be a benefit as existential issues related to AGI remain unresolved. Think of the alignment problem and consequent doomsday scenarios. Current AI is good enough for most practical problems in many specialized domains. We have found a wild horse. Now we need to saddle it.
In 2026, I expect many commercial enterprise AI products. Most of them will disappear as quickly as they pop up. Some will remain and blend into everyday life, just like products from the past. Few people today label search engines or Apple’s Siri as AI products, even though both clearly are.
The year 2026 will therefore not be about which organization has the best LLM. It will be about how effectively existing models are used in product development to deliver real value to users.
Social Media Shift
Beyond enterprise products, I also expect a shift in social media and interpersonal communication. AI slop, meaning automatically generated spam on platforms such as LinkedIn or Facebook, is a real problem that needs to be addressed. Unmoderated channels are flooded with misinformation and trivial content. High-quality posts, if they appear at all, are lost in a sea of AI slop.
Algorithm changes and stricter moderation are necessary to counter the dark predictions of the dead internet theory. In that scenario, only bots communicate with other bots online; humans left the internet completely. The social internet will likely become more private. People will limit communication to circles of real friends. Social platforms will increasingly resemble media outlets. More high-quality content will be labeled and paid for. This is not necessarily bad. Quality content and journalism have real costs and cannot be free. On the other hand, this creates risks of centralization, restrictions on free expression, and manipulation of public opinion.
The last mentioned is an issue that could easily spiral out of control in 2026. Sam Altman has stated that many young people do not make major decisions without consulting a chatbot. That level of trust can be exploited, both commercially and politically.
Finally, there is the issue of social isolation among young people. This trend started during the pandemic and was amplified by technology. In a time of aging populations across most developed countries, this is a problem that should not be underestimated.
And Not Much Else
Besides AI, can we expect a new software development paradigm comparable to microservices? The industry is currently fully occupied with understanding how to use AI effectively without losing control. Technologies such as agentic AI and MCP are still in their infancy. I don’t expect the year 2026 to be enough for their full stabilization. Given the scope of this challenge, there will be little capacity for a parallel revolution.
This situation is not new. Truly novel ideas have been rare in recent years. Even the most influential figures in the field, such as Martin Fowler and Kent Beck, focus mainly on consolidating existing methods and explaining established best practices. Software development is gradually shifting away from being a scientific discipline driven by experimentation and is becoming more like pure engineering. Work follows general rules defined for a given context, with little novelty.
Frankly, there is no strong reason for otherwise. The current ecosystem already provides sufficient tools to build software that meets virtually any contemporary requirement. It’s the AI systems are the new kids on the block, pushing our 2025 technology to its limits and representing the main challenge for 2026.
Happy new year!