Loading video player...
What will be the most important trends in AI in 2026? Well, we take a stab at this every year
with with some success, I would say. And this time out, I have the knowledgeable assistance
of my colleague, Aaron Baughman, to help us out. Well, yeah. You know, after your prediction
of infinite memory last year, I thought maybe you could use just a little bit of help. Yeah,
that's that's fair. Well, how about we each take four trends each? That sounds good. How about you
first? All right. Okay. So my number one trend of 2026 is multi-agent orchestration. Now last
year we said 2025 was the year of the agent. AI agents that can reason and plan and take
action on a task and agents I think it's fair to say really delivered. There are new numerous
agentic platforms for tasks like coding and basic computer use but no single agent really excels at
everything. So, what if you had a whole team of agents working together? So, maybe we've got an
agent here that kind of acts as a planner agent that decomposes goals into steps. Maybe we have
some worker agents here that do different steps like one specializes in writing code,
others call APIs and so forth. And then perhaps we have a critic agent that evaluates outputs and
flags issues. And these agents collaborate under a coordinating layer that is the orchestrator.
And multi-agent setups like this help introduce cross-checking where one agent checks the other
agents work and it can break problems into more discrete verifiable steps. Well, great. So,
how could I really follow that trend? Well, I think I might just have one. So,
the second one is going to be the digital labor workforce. So now these are digital workers that
are autonomous agents that can do a couple of items. So the first one is they can parse
a task by interpreting multimodal input. So after preparation the worker then executes what's called
a workflow. Now this is where at the end of an action plan you know it would follow a sequence of
steps but then it has to be integrated into some sort of system that then in turn can take action.
And these could be downstream components. Now these systems are then further enhanced by what we
call human-in-the-loop AI, which then provides a couple of items. The first one would be oversight.
The next one would be correction and then we're looking at these strategic guidance or these rails
um to ensure that all of these agents are doing what they're supposed to be doing. Now this
overall trend will create a force multiplying effect to extend human capability. Now trend
number three is physical AI. Now we all know that large language models they generate text like ABC.
And then there are other models as well. So for example there are plenty of diffusion image models
and they generate pixels. They generate images. These are all operating in digital space. Now,
physical AI is about models that understand and interact with the world that we live in, the the
real 3D world. And this is about models that can perceive their environment, reason about physics,
and that can take physical action like robotics. So, previously getting a robot like this to do
something useful meant programming explicit rules. So if you see an obstacle, you should turn left,
for example. And it was all done by humans. It was up to yeah, smart guys like this to code these
rules. Now, physical AI kind of flips that around. So you train models in simulation that simulate
the real world and it learns to understand how objects behave in the physical world,
how gravity works, how to grasp something without crushing it. Now these models are sometimes called
world foundation models. They're generative models that can create and understand 3D environments.
They can predict what happens next in a physical scene. And in 2026, many of these world models are
taking things like those humanoid robots that you found there, Aaron, and they're taking them from
research to commercial production. Physical AI is scaling. Well, Martin, you just took my trend,
but let's just go ahead and say number four is about social computing. Now, this is a world
where many agents and humans operate within the shared AI fabric. So say if I have an agent here
and then a human here. So they're going to be connected through this fabric and here if I
have information that flows between the two, they begin to understand each other and then they can
gather what the intent is going to be. And then once they have the intent and information, they
have actions. They can affect each other or maybe even the environment of which they're in. But all
of this flows seamlessly across this system. It's this shared space that enables collaboration,
context exchange as well as event effective understanding. Now the outcome is really an
empathetic emergent network of these interactions. It's what we call this collective intelligence
or this real world swarm computing. So teams of agents, digital labor, humanoid robots, and tech
that can understand me with effective computing. 2026 could be uh quite the year and we're only
halfway through the trends. So trend number five that is verifiable AI. Now the EU AI act is coming
and by mid 2026 it becomes fully applicable. And think of this a little bit like GDPR but for
artificial intelligence. Now, the core idea here is that AI systems, especially high-risk ones,
need to be auditable and they also need to be traceable. Now, what does that mean? Well,
it means a few things. It means documentation. So, if you're building high-risk AI, you need
technical docs that demonstrate compliance to how you tested the models and the risks that you
identified. It means transparency. So, users need to know when they're interacting with the machine.
So things like synthetic text, they need to be clearly labeled and it means data lineage. You
need to be able to summarize where your training data came from and prove you respected copyright
optouts. And just like how GDPR has shaped global privacy, not just folks in the EU, the EU AI act
will probably set the template for AI governance worldwide. Wow, that's great. And you know, trend
number six, right? It really changes everything, but it also changes nothing at the same time.
And now this is where we put in quantum utility everywhere. So 2026 is where we start to see this
quantum computing to reliably start solving real world problems better, faster, or more
efficiently than classical computing methods. Now, at this point, we have this quantum utility scale.
is these systems that begin working alongside and together with classical infrastructure to deliver
these practical value in everyday workflows. Now, this is going to help with optimization and then
we'll also look at simulation and decision-making. Now, all three of these tasks were previously
out of reach within the classical realm. But this hybrid quantum classical error, it will begin to
transform quantum computing into this mainstream paradigm as it's going to be woven into our
everyday business operations. Now my trend number seven is reasoning at the edge. Now last year, we
talked about very small models, models with just a few billion parameters that don't need huge
data centers to run. They work on your laptop or well maybe even your phone. Well, in 2026,
those small models are learning to think. So, if we think about the best models that we have today,
the frontier models, well, pretty much all of them now use something called inference time compute.
They spend extra time thinking before giving you an answer, working through problems step by step.
Now, the trade-off for that is they need more compute. But here's what's changing. Essentially,
teams have figured out how they can distill all of this reasoning information into smaller models.
So now these smaller models can perform thinking as well. You're taking massive reasoning models
that generate tons of step-by-step solutions and we're using that data to train the smaller models
to reason the same way. And that's resulting in reasoning models with only a few billion
parameters. They work offline. Your data never leaves your device. And there's no roundtrip
latency to a data center. So for anything that's real time or mission critical, having a model that
can actually reason through a problem locally is a pretty big deal. Yeah. So that's all very true,
Martin. But now our last and final trend is number eight. So this is what we're calling amorphous
hybrid computing. So this is a future where both AI model topologies and the cloud infrastructure,
they blend into what's called a fluid computing backbone. So AI models, they're shifting beyond
just this pure transformer design, right? They're beginning to evolve into these other architectures
that integrate transformers and we call them these state space models. And then in 2026,
you're also going to see different emerging algorithms that are combine both the state space
and transformers and other elements together, right? And that's going to be really fun to watch,
very artful. And then at the same time, we have this cloud computing piece that's becoming fully
differentiated by combining many different chip types. So we're going to have CPUs,
GPUs, TPUs as well. And finally, what we just talked about in trend six, quantum, we're going
to have QPUs. I did also want to mention and note that you'll see these neuromorphic chips
that are coming out and those emulate the brain. But all of these are going to be put together
right into this unified compute environment where parts of each of these types of models,
they're going to be automatically mapped to the optimal compute substrate. And this is
really going to help to deliver this maximum performance and efficiency. And you know what?
Who knows? But at this pace, probably not in 2026, but I think further out,
you might see DNA computing entering into the mix. Well, those are some lofty goals. And look,
these are what we think are some of the biggest AI trends in 2026. But what are we missing?
Which AI trend do you expect to be a big deal in 2026? Yeah, let us know in the comments below.
Ready to become a certified watsonx AI Assistant Engineer v1 - Professional? Register now and use code IBMTechYT20 for 20% off of your exam → https://ibm.biz/BdbTDQ Learn more about AI Trends Shaping the Next 10 Years here → https://ibm.biz/BdbTDT What will define AI in 2026? 🚀 Martin Keen & Aaron Baughman explore groundbreaking trends like Agentic AI, cloud computing, automation, and quantum computing, plus innovations like Physical AI. Discover how these technologies will transform industries and drive the next wave of intelligence. AI news moves fast. Sign up for a monthly newsletter for AI updates from IBM → https://ibm.biz/BdbTDk #agenticai #cloudcomputing #automation