In the technological world we are fascinated with chatbots who write essays and take their time reasoning. Grok 3 goes, Claude 3.7 It comes, meanwhile, something less visible but deeper is happening: The beginning of the merger between conversational and mechanical bodies.
For the first time, robots not only execute preprogrammed instructions. Now also, in their own way, they understand.
Historically, robotics and AI have followed separate paths. Parallel, but separate. Industrial robots were as accurate as stupid. AI systems are intelligent, but incorporeal.
Let’s think about robotic arms that have existed on assembly lines for decades. Millimetrically exact, but absolutely lost If a single component appeared in a position slightly different from that expected.
The new generation of robots connected to LLMS You can now interpret ambiguous instructions, such as “bring me something for thirst”, and solve the problem through reasoning (word of the year), evaluating which drinks are available, if the user showed preference for some and even if there is ice in the freezer.
We no longer program specific movements, but rather general objectives.
Figure robots are good examples. So good that even They work autonomously at a BMW factory. According to The company has just publishedthey can even receive generic verbal instructions, such as collecting pieces, and without the need for previous specific programming are able to visually analyze the environment and detect them.
They can even pause, reassess the situation and correct the error if someone modifies the pieces. This contextual adaptation capacity was unthinkable a couple of years ago.
The really groundbreaking of this impaired in robots is that You can learn very differently. The LLMS trained with text lack the physical understanding of the world. Traditional robots lack contextual intuition. By merging them an intelligence that includes both semantics and physics.
A robot equipped with LLMS is not only able to understand the instruction “opens that box without damaging its content”, but can improvise before boxes never seen, evaluating materials, closures and fragilities.
The revolution, unfortunately, will not be spectacular as in The novelty-fictionbut it will arrive in the form of robotic arms in factories that can be reconfigured with a verbal order. Or warehouse robots that will understand contextual priorities. Or medical assistants capable of interpreting non -verbalized needs of their patients.
Boston Dynamics, the Non-Va-Más de la Robotica during this last decade thanks to your robots jumping and doing Parkourshe is no longer as interested in acrobatics as in integrating understanding systems that allow her machines Understand complex instructions in construction and industry environments. You just have to see Your website. And on the horizon look The Tesla optimus or the Cyberone de Xiaomi. Or unitree like One of the great Chinese technological bets.
The big change will come When these systems stop failing before the unforeseen and begin to apply general principles of physical and contextual reasoning. We are not seeing the birth of artificial consciousness, but the understanding of the physical world and the world of meaning in a single integrated system.
What this powerful convergence does is its silent nature. He catches us arguing On whether Grok 3 deserves a better product or about itself Chatgpt 4.5 It will be sufficient during the remaining of the year, but Robots are beginning to understand the world like us. Not only by calculating a trajectory, but understanding intentions, contexts and meanings.
That is much more transformative and valuable than Any ten -page trial generated in seven minutes.
Outstanding image | Figure, Ryunosuke Kikuno in Unspash
GIPHY App Key not set. Please check settings