AI and Driving: The Next Chapter of Distracted Driving

For years, the public message about distracted driving was easy to visualize.

A driver held a phone. Then drivers started texting. Then came social media, navigation systems, touchscreens, Bluetooth and voice assistants.

Each generation of technology changed the way distraction looked.

Now we are entering another stage.

A person can place the phone in a mount, connect it to the car, keep both hands on the steering wheel, keep their eyes facing the road—and still be deeply engaged in a conversation that has very little to do with driving.

That conversation may be with artificial intelligence.

This is not an argument against AI. It is not an argument against smartphones, texting, social media, navigation or phone calls as technologies. All of these tools can be useful. The problem begins when a useful activity competes with another activity that requires continuous attention.

Driving is one of those activities.

The evolution of distraction

Distracted driving did not begin with AI, and it did not begin with smartphones.

Drivers have always been capable of becoming distracted: by passengers, radios, food, maps, thoughts, arguments, fatigue, roadside events or the simple tendency of the mind to wander.

Technology did not invent distraction. It multiplied the number of things competing for attention.

The sequence is familiar.

Phone calls.

Text messages.

Hands-free Bluetooth.

Social-media notifications and scrolling.

Voice assistants and in-vehicle information systems.

And now:

Conversational AI.

Each step can create the impression that the previous problem has been solved. A headset solved the need to hold the phone. Bluetooth solved the cable. Voice control reduced the need to touch a screen. A conversational AI interface can go even further: the entire interaction can happen by voice.

From a mechanical point of view, that can look safer.

But road safety is not only mechanical.

Hands-free is not mind-free

Safety research commonly separates distraction into three broad forms: visual distraction, manual distraction and cognitive distraction. The first two are easy to understand. If your eyes leave the road, you may miss something. If your hands leave the controls, your ability to respond may be reduced.

Cognitive distraction is less visible.

The driver may appear to be doing everything correctly. Eyes forward. Hands on the wheel. Body positioned normally.

But attention is somewhere else.

The U.S. Centers for Disease Control and Prevention describes cognitive distraction as taking the mind away from driving. NHTSA similarly defines distracted driving broadly as activity that diverts attention from the task of safe driving. Research from the AAA Foundation for Traffic Safety has examined the mental workload of hands-free conversations, speech-to-text systems and voice-based assistants and found that these interactions can impair aspects of driving performance even when the driver is not physically holding a device.

That matters because a technology can reduce one form of distraction while leaving another intact.

Hands-free can solve the problem of the hand.

It cannot automatically solve the problem of the mind.

A conversation is not just sound

We often talk about a phone conversation as if the only thing happening is speaking and listening.

That is not how the brain experiences a demanding conversation.

Conversation can require memory, prediction, language, judgment, emotional processing, decision-making and the continuous construction of what to say next.

A casual exchange—“How are you? I am almost there. I will call you later.”—may require relatively little mental work compared with a discussion in which somebody is trying to solve a complicated problem.

The exact level of distraction varies with the person, the road, the traffic, the subject and the complexity of the task. A phone call can also be highly distracting. A passenger conversation can become intense. There is no universal rule that one conversation format is always safe and another is always unsafe.

But complexity matters.

Now consider how many people use conversational AI.

They are not always asking for the weather.

They may be developing a business idea. Troubleshooting a machine. Comparing financial options. Planning a project. Arguing through a technical question. Dictating a document. Asking the system to critique an idea. Listening to a long answer. Correcting it. Reframing the prompt. Remembering an earlier point. Evaluating whether the answer is credible. Deciding what to ask next.

That is not simply audio playing in the background.

It is work.

Why conversational AI deserves special attention

Modern conversational AI has an unusual characteristic: it encourages continuation.

A traditional voice command often has a narrow endpoint.

“Call home.”

“Play this song.”

“Navigate to this address.”

The user gives the command, the system performs an action and the interaction ends.

Conversational AI can be different.

One answer creates another question. The system may ask for clarification. The user may disagree. A new idea appears. The discussion branches. Ten minutes can pass without the interaction feeling like a single continuous task.

The driver can become mentally invested in finishing the thought.

That is precisely what makes this form of distraction difficult to recognize. There may be no obvious moment when the person thinks: I am now doing a second task.

It feels like talking.

But depending on the topic, it may actually be problem-solving.

The dangerous illusion of looking at the road

One of the most important lessons from cognitive-distraction research is that looking toward something is not the same as fully processing it.

A driver may be visually oriented toward the road but mentally occupied by another task. That creates a dangerous illusion: because the driver remembers looking forward, they may believe they were fully attentive.

This is one reason cognitive distraction deserves more public attention. Manual and visual distraction leave evidence. The phone is in the hand. The eyes are directed at a screen.

Mental distraction leaves almost nothing visible.

There is no warning light that says: your brain is currently allocating too much attention to this conversation.

This is not an anti-AI argument

It is important to separate criticism of a use case from criticism of a technology.

AI can be extremely useful before a trip, after a trip or while a vehicle is safely parked. It can help plan a route, organize work, summarize information, prepare questions, draft notes or help solve a technical problem.

The same principle applies to many other technologies.

Texting is not bad because written communication is bad.

Social media is not bad because sharing information is bad.

Phone calls are not bad because conversation is bad.

The issue is context.

There are activities that are perfectly reasonable at a desk and irresponsible behind the wheel.

The tool does not have to be dangerous for the timing to be dangerous.

Research gives us a warning, not yet a complete answer

There is an important scientific distinction to make.

Existing research has directly examined cognitive distraction from phone conversations, speech-to-text systems, in-vehicle information systems and earlier generations of voice assistants. That research provides strong reason to take mentally demanding hands-free interactions seriously.

Modern generative conversational AI is newer and can support much longer, more open-ended and more cognitively complex exchanges. It would therefore be irresponsible to pretend that every possible AI-driving interaction has already been measured and ranked.

The correct conclusion is more modest—and more useful.

We already know that mental workload matters behind the wheel. We already know that voice-based systems can create meaningful cognitive demand. When a new technology makes it easy to conduct extended problem-solving conversations while driving, that combination deserves caution and further research.

The practical rule is simple

If an AI conversation starts requiring real thought, it should stop being a driving activity.

That means the moment you find yourself trying to analyze, compare, calculate, debate, compose, troubleshoot, plan or make a significant decision, the safest option is to end the interaction and return to it when the vehicle is safely parked.

Not at the next interesting part of the conversation.

Not after one more question.

When it becomes mentally demanding.

Driving should not become a mobile conference room, engineering office, therapy session, strategy meeting or brainstorming workshop simply because voice technology makes those activities technically possible.

AI systems can help too

Responsibility does not belong only to drivers. Designers of AI products, mobile operating systems and vehicle interfaces can also reduce risk.

A system that knows it is being used through a vehicle interface could favor brief, low-demand interactions. It could discourage long analytical exchanges while the vehicle is moving. It could offer to save the discussion and resume later. It could make “continue when parked” a normal part of the experience instead of treating interruption as failure.

Good design does not only ask, Can the user do this?

It also asks, When should the system make it easier not to?

The next chapter of distracted driving

For years, distracted-driving campaigns focused on the visible device.

Do not hold the phone.

Do not text.

Do not look down.

Those messages remain important.

But the next chapter may require an additional message:

Do not give a conversation the part of your mind that the road needs.

Conversational AI is powerful precisely because it can hold our attention, respond intelligently, follow a complex thread and help us think.

Those are strengths.

Behind the wheel, the same strengths can become a competing demand.

The future of road safety will therefore depend not only on where our hands are and where our eyes are.

It will depend on recognizing where our thinking has gone.

Some conversations can wait.

Some ideas can wait.

Some questions can wait.

The road cannot.