Your TV’s ‘AI Picture Mode’ Isn’t as Smart as You Think

Tutorials

Your TV’s ‘AI Picture Mode’ Isn’t as Smart as You Think

Following the explosion in popularity of services like ChatGPT, the term “AI” was suddenly in everything. Much like the dot com boom and bursting bubble, simply adding the word “AI” to your company or product could boost sales and stock prices.

TVs were no different, and there are plenty of models with “AI” in the name, or prominently somewhere in the feature list. The most common is some sort of “AI picture mode”. But if you’re picturing some sort of smart electronic brain processing your video, that’s not quite the truth.

The Rise of the ‘AI Picture Mode’

Getting the best picture quality on a TV has been a manual process for most of the history of these devices to date. However, as TVs became “smart” and could run more sophisticated software using more powerful processors, that started to change.

By activating this “smart” or “intelligent” picture mode, the TV will select the right brightness, sharpness, contrast, and whatever else for the image and even the environment. Thanks to ambient light sensors in some TVs. Of course, this predates the use of “AI” in TV marketing and model names, so what’s actually going on under the hood?

What “AI” Actually Means Inside Your TV

One problem here is that “AI” is a pretty broad term. It’s a field that includes so many different technologies, techniques, and concepts that it’s not super-useful as an everyday term. What people think when they hear “AI” today is something sophisticated and almost human-like in intelligence, but what’s in your TV is just an algorithm or a heuristic decision-making system that’s not necessarily any more advanced than before the “AI” sticker was slapped on the box.

These systems work at best like technologies such as NVIDIA’s DLSS, where lots of machine learning training was done on sample footage, and the end result is a lightweight algorithm that can run on the hardware on a typical consumer TV. It’s not like there’s any “thinking” or adaptive learning going on.

Why the Results Often Look Worse, Not Better

You may not know this, but often the actual display panels on an expensive brand-name TV and a cheaper TV from a less well-known brand are actually the same panel. There are only so many companies that make LCD and OLED panels, and they sell them to other companies in addition to making their own TVs. So why do some of those TVs end up looking so much better if the screens are essentially the same?

The answer is image processing, which is more often than not the divider between mediocre image quality and motion handling, and an exceptional experience. I bring this up, because these different “AI” technologies can vary widely in how well they do the job.

For example, AI upscaling on one brand or model of TV can be brilliant, producing images that look very much like the targeted resolution should. On other TVs, you get weird artifacts, motion glitches, or details destroyed by a poorly-made algorithm.

These “AI” modes are mostly used to handle older content or lower-resolution content that doesn’t have the metadata of modern video. For example, if you’re watching something mastered in Dolby Vision, and your TV supports Dolby Vision, it knows exactly how the scenes are meant to be presented. No need for fancy AI.

The Marketing Problem: AI as a Buzzword

These new “AI” features are largely just the old “Dynamic Picture Mode” or “Intelligent Picture Mode” features that have been around for some time. Of course the technology has become better, and can offer amazing results, but it’s still just a marketing buzzword that’s meant to make a product sound more premium or advanced than it really is.

It’s like the words “nano” or “quantum” being used everywhere at one point, because it sounds futuristic.

How to Actually Get a Smarter Picture

A Sony Trinitron CRT TV with sharpness set to zero.

The good news is that you don’t have to be swayed by AI picture modes, and that you can get a good image without resorting to imperfect automation. The optimal answer is to get a professional to calibrate your TV, or attempt it yourself using a guide and the right equipment. I’m also a big fan of using Filmmaker Mode, which basically switches off most of the worst processing, and then use that as a basis to just tweak the image to my taste.

It’s also a good idea to do a little research into AI features of a TV to see if they actually use machine learning technology. For example, LG’s AI Super Resolution feature actually uses machine learning hardware to upscale images intelligently, which is about as close to a true AI feature as you’re likely to get on a TV right now.

Most importantly, when looking for a new TV, focus on the fundamentals of good picture quality, and image processing. Gimmicks are always the bane of consumer AV tech, but a TV that gets all the fundamentals right will still keep you happy long after the current marketing fad has passed.