Self-aware cars: how AI will transform vehicles by 2036
Artificial intelligence promises cars far more than just voice assistants or automatic parking. Over the next decade, it could transform driving, maintenance, operating costs, and even the criteria for choosing a used car. Yet fully autonomous mobility will remain the most challenging part of this transformation.
Imagine a car that assesses its battery condition before a trip, suggests the most cost-effective route, detects signs of malfunction, and explains why they should be checked. On the road, it helps avoid dangers, and upon return, it recommends a time for maintenance.
Individual components of such a scenario already exist. The next step is to integrate them into a system that understands the context of the journey and the owner’s needs.
By 2036, AI is likely to transform mass-market vehicles primarily due to improvements in safety, diagnostics, and usability. The adoption of cars capable of driving autonomously on any road will be slower and more uneven. Below is an analysis of existing technologies along with editorial forecasts for their development.
The car will begin to better understand road conditions
Current assistance systems are not sufficient merely to detect pedestrians or measure the distance to nearby vehicles. They must also assess what might happen next: whether a person will enter the road, if a car will start to change lanes, or if a parked van poses a hidden danger.
Predicting the behavior of other road users will become one of the key areas of development for automotive AI.
A notable example is the Alpamayo models and simulation tools unveiled by NVIDIA in 2026. They are designed to advance autonomous driving, including training for complex and rare road scenarios. These are tools for developers, not proof that any vehicle is ready for self-driving operation.
For end-users, the practical benefits may be simpler: fewer false warnings, more accurate emergency braking, smoother operation of adaptive cruise control, and more reliable assistance at intersections.
At the same time, even a powerful model is limited by the quality of information available. Dirty cameras, snow, glare from the sun, or unusual traffic patterns will remain problems. AI does not eliminate the need for reliable sensors and proven safety mechanisms.
Autopilot will continue to evolve, but “driverless car” will have different meanings
The biggest mistake is to consider any system that turns the steering wheel and applies brakes as a fully autonomous driver.
For example, Mercedes-Benz describes MB.DRIVE ASSIST PRO as a system that combines navigation with assistance during urban driving. However, it belongs to level 2 automation, which requires constant driver supervision.
The difference between the main levels is fundamental:
Level
What it means for the driver
Level 2
The system assists with steering, but the driver must continuously monitor traffic
Level 3
Under specific conditions, the system handles driving tasks; the driver must be ready to take over at any time
Level 4
Within permitted conditions and areas, the system can operate without a driver.
Level 5
Automation covers all road conditions that a human can handle.
This classification is used by the U.S. agency NHTSA. It allows for distinguishing between driver assistance and fully autonomous driving. (nhtsa.gov)
Over the next decade, it is logical to expect expansion in ride-hailing services, autonomous transportation on designated routes, and systems for highways. Widespread use of driverless private vehicles on any roads is a much less certain scenario.
The reason lies not only in computing power. It is necessary to prove safety, define operating conditions, and establish rules for liability. A complex yard with children, snow, and temporary barriers can be a more challenging task than hundreds of kilometers of highway.
The voice assistant will evolve into a trip organizer
In place of a set of commands, the driver can express a need: to find a charging station along the way, select a lunch stop, or ask for an explanation of a message on the dashboard.
The direction this is heading can be seen in the partnership between Mercedes-Benz and Google Cloud. Their automotive assistant is evolving toward multi-step conversations while maintaining context of the discussion.
By 2036, such systems may integrate navigation, calendars, charging, and service functions. However, their usefulness will depend on data accuracy: a convincing response about road conditions or malfunctions should not replace verified information.
Critical commands will require separate oversight. The voice assistant can suggest an action, but a freely generated response should not directly control the brakes or steering.
Service to Become More Personalized
This will also change how auto repair shops operate: alongside mechanical diagnostics, data analysis, sensor calibration, and software checks will become more important.
Electric vehicles will benefit more from the same battery
AI cannot programmatically increase a battery’s physical capacity. However, it can help assess charge levels more accurately, predict degradation, and determine optimal operating modes.
Bosch is already developing Battery in the Cloud—a solution for analyzing battery condition and predicting its lifespan. The company also mentions using battery health data to estimate a vehicle’s residual value.
The coming decades may bring more accurate predictions of range and personalized charging recommendations. The potential benefit lies in better utilization of existing resources and reduced degradation.
For used electric vehicle buyers, a high-quality battery report might become more important than the mileage figure. However, trust in such reports will require transparent methodologies and protection against manipulation.
AI Will Transform Cars Even Before They Arrive in Showrooms
Some of these changes will go unnoticed by drivers: artificial intelligence will be used in design, testing, and production monitoring.
BMW, for example, employs AI models to analyze the relationship between battery cell production parameters and the resulting outcomes. The goal is to make processes more stable and ensure better quality control.
In the long term, this could shorten some development cycles and reduce manufacturing defects. However, cheaper development does not guarantee a cheaper car: savings may be offset by new sensors, powerful computers, and costs related to software support.
You’ll have to pay for intelligence — even after purchase
The advancement of AI will increase a car’s dependence on software. For buyers, not only specifications and warranty but also update frequency, availability of services in their country, and the ability to transfer functions to the next owner will become important.
The risk is that a mechanically sound vehicle might lose some of its digital capabilities once support ends.
A separate issue is personal data. Routes, voice commands, and information about driving style can be useful for services, but owners need to understand what is being collected and to whom it is transferred.
Cybersecurity and update management are already covered by separate UN regulations — R155 and R156. Thus, software support becomes part of vehicle safety.
What the Next Decade Could Hold
This is an editorial scenario for development, not a confirmed timeline from manufacturers:
Period
Most Likely Changes
2026–2029
Wider adoption of voice assistants, improved driver assistance, battery analysis, and service data
2030–2033
Deeper integration of vehicles with charging and service platforms; expanded autonomous driving in permitted conditions
2034–2036
Rising role of software support in vehicle value; further expansion of autonomous services in designated areas
For Ukraine, more accurate diagnostics, assessments of the condition of used vehicles, Ukrainian-language assistants, and the ability to operate key functions without constant internet access could prove most beneficial. The possibility of using the system after importing a car will be just as important as its presence in the factory setup.
The main change by 2036 is that cars will increasingly be evaluated based on the quality of their decision-making. How accurately they warn of dangers, predict costs, explain malfunctions, and maintain their capabilities over time. It is these qualities that can offer owners greater everyday benefits than the promise of completely eliminating the need for a steering wheel.