Artificial Intelligence and Robotics

Artificial intelligence is the field of building computer systems that can do things we would normally say require intelligence: perceiving the world, learning from experience, reasoning, making decisions, and using language. John McCarthy, who coined the term, described AI as the science and engineering of making intelligent machines, especially intelligent computer programs, and defined intelligence itself as the computational part of the ability to achieve goals in the world. An important point in his definition is that AI does not have to copy the way the human brain works; it can use any method that gets results. Put simply, AI is software that takes in information, works out what is going on, and chooses actions that move it toward a goal, ideally getting better at this as it gains more data or experience.

• Machine learning: Algorithms that learn patterns from data instead of following rules written by hand. Deep learning, which uses neural networks with many layers, is the branch behind most of today’s progress.
• Computer vision: Teaching machines to understand images and video, for example recognizing objects or faces and tracking movement. It is used in self-driving cars, medical imaging and security cameras.
• Natural language processing (NLP): Getting computers to understand and generate human language. Translation, chatbots and large language models all come from this field.
• Robotics: Combining perception, planning and control so that machines can act in the physical world, not just process information.

AI as a field started in the mid-1950s, but for decades progress came in waves, and periods of hype were followed by “AI winters” when results fell short. The current boom comes from several factors arriving at the same time.

The first is data. The internet, smartphones and sensors now produce enormous datasets for models to learn from.

The second is computing power. GPUs, originally built for video games, turned out to be ideal for training neural networks. According to Stanford’s 2026 AI Index, global AI compute capacity has grown about 3.3 times per year since 2022, reaching the equivalent of roughly 17.1 million top-end Nvidia H100 GPUs.

The third is better algorithms. Deep learning took off around 2012 when neural networks started winning image-recognition competitions, and the transformer architecture introduced in 2017 led directly to today’s large language models.

The last factor is that AI now works well enough for everyday use, which creates a cycle of more users, more investment and more research. Generative AI reached 53% population adoption within three years, faster than the PC or the internet did. AI models now match or beat human baselines on PhD-level science questions, competition-level mathematics, and multimodal reasoning.

• Manufacturing (especially automotive and electronics): Robot arms weld, paint, assemble and package products. The International Federation of Robotics reported that 542,000 industrial robots were installed worldwide in 2024, the second-highest yearly count in history. In the US, the automotive industry remains the largest adopter. The reasons are precision, consistency, the ability to work around the clock, and taking over tasks that are heavy, repetitive or dangerous for people.
• Transportation and logistics (warehouses): Autonomous mobile robots move, sort and handle goods. In 2024, 102,900 transport and logistics robots were sold, a 14% increase, and they made up roughly half of all professional service robots. The main driver behind this growth has been labor shortages, along with the huge volume of e-commerce orders.
• Healthcare and medicine: Surgical robots help doctors perform precise, minimally invasive operations, and other robots deliver supplies inside hospitals. Sales of medical robots rose 91% to about 16,700 units in 2024, as a growing elderly population increases demand. Robots here offer precision, reduce strain on staff, and can lower infection risk.

• Navigating a huge, crowded and changing space: GPS does not work indoors, so the robot has to build its own map and continuously work out where it is, usually with lidar and cameras. A mall makes this harder. Glass storefronts and shiny floors confuse sensors, the layout changes with kiosks and seasonal decorations, and the robot must move safely around crowds, children, strollers and escalator edges. Because it is wheeled, it cannot use stairs, so covering multiple floors means using elevators or deploying a separate robot on each floor.
• Staying powered for 24 hours: No battery lasts a full day of patrolling, so the robot must track its charge, return to a docking station on its own, line up correctly and recharge, all without leaving gaps in coverage. Real systems handle this in different ways. Germany’s SentryBot reaches its recharging station without human help, which is what makes 24-hour operation possible. Security robots in Tokyo’s Metropolitan Government Building automatically look for charging ports when their battery runs low and keep filming while they charge.
• Recognizing real threats with few false alarms: The robot has to tell normal activity from suspicious activity. During the day that means thousands of shoppers, and at night it means a dark, empty building. Lighting and context change completely between the two. A cleaner at 3 AM should not set off an alarm, but an unattended bag or smoke should. It also needs a reliable network connection across the whole building to stream video to the control room. Recording shoppers raises privacy concerns, and the robot has to withstand tampering.