Neural networks: 5 ways they will change your life forever

And what can go wrong with each one

We live in the strangest of all possible futures that anyone ever predicted. With almost a quarter of the XXI century down, and we still have no flying cars or interplanetary spaceships. 30 years ago, who would’ve thought that big data, supercomputers and social networks will dominate the technological horizon in 2021?

The name of the future is neural networks (NN), a bunch of algorithms that imitates the operations of a human brain to recognize relationships between vast amounts of data. Through the great power of meme culture, we all are aware of what NN are able to do: singing gifs, remastered graphics and videos, old photos alive again. Cool stuff, no doubt that we needed it.

But NN’s ability to process data is so universal that it will undoubtedly affect some things we consider to be our own, human jobs. Now they can replace us, whether we like it or not.

Neural network and cinema

How it works

Literally every face that ever existed in the graphic form (from medieval portraits to modern selfies) now can be animated, reanimated and returned to the screen. It has been a long way since 2004’s sci-fi movie «Sky Captain and the World of Tomorrow» where «starred» Laurence Olivier, who died in 1989. Now CGI studios are literally able to reconstruct any actors in any age and use them in movies in any way they want.

What can go wrong

Deepfakes (using the open code of neural networks to manipulate the existing footage of a person) were considered such a big threat that Facebook specifically forbade them as a type of fake news. Google (who ironically, owns the most powerful NN in the world) is currently developing the methods of guaranteed deepfakes’ detection. With social networks as primary reality of today, misinformation of political leaders saying some fake things can lead to uncontrolled consequences. Or this could be you, on videotape, «doing» some obscene or even criminal things.

That’s the technological price for such an amazing invention. But there are also economical and ethical consequences, for example, for acting as a profession. Actors are still the most expensive part of making a movie. Why pay actors or respect their rights if you can replace them with digital copy? Why pay them each time when you can make them sign a contract of using their face and voice to put it on stunt doubles (who are doing all tricks anyway). That idea of selling a digital scanning instead of actual acting was explored in another movie, «The Congress» (2013), before the NN phenomenon exploded and it seems very real already.

What can you do

Be critical to the videos you see online. Especially to the ones that demand an instant emotional reaction from you. For example, political-themed or compromising ones – anything can be taken out of context or strictly deepfaked, so you need to think before reacting.

Neural Network Structure
Neural Network Structure

Neural network and programming

How it works

For many years (basically since the 1980s) «IT guys» were the elite, the workers who literally created and managed the framework for our world. Programmers were (and still are) treated like priests of unknown religion because not a lot of people can understand what they are actually doing. That’s the reason why they are paid SO much. But as it turns out, NN can do this too, at least on Junior level.

New project from Microsoft and Open AI, GitHub Copilot is the «adviser» that can analyze the existing code and suggest the improvements or even completes it. I.e. Copilot does the job of Junior programmers and sometime in the near future – as it will learn enough, studying the existing types of code – it will write the code itself.

Right now, the programmers, charmed by the new tool, are “teaching” it by approving or discarding Copilot suggestions how to improve the code. It is already good enough with handling JavaScript, Python and TypeScript.

What can go wrong

There will be less basic-level programmers in the future, they will be replaced by AI (artificial intelligence, whose foundation the neural networks are), capable of solving problems considered impossible or very difficult by human standards. Also: it was already not rocket science to write a computer virus. With the help of AI this will be as easy as playing a video game. Thousands of new cyber threats will flood the web, constructed from ready “bricks” of code, altered and “improved” with a NN little helper.

What can you do

Always use the antivirus. It won’t be a problem if neural network heuristic analysis will be implemented by the devops to detect such custom-made threats. You will fight NNs with NNs.

Neural network and artists

How it works

Of course, the individual artists will always be demanded. But there is nothing magical in the very process of organizing the “correct” sequences of notes or color patterns. NN can do this even now. Neural networks are already able to “improve” the existing content: making mono a stereo, making low-res video into hi-res (the most stunning example is classic 1896 short film «Arrival of a Train» that was upscaled to 4K by NN), and so on. The computer can’t become a Mozart, but it can generate an endless stream of ambient or upbeat music or a «typical» classical tune for the background. It can’t become a Van Gogh but it can transform the photo of your dog into the Van Gogh-stylized painting of your dog that is very convincing.

With the development of 3D printing, NN will also replace the sculptors. It already proved to be useful for archaeologists: they put physical discovered objects into digital form through 3D scanning, and neural networks tell what the missing elements should look like. Same with design (interiors, clothing or decor): all can be translated into code and improved.

The art is a number of very complex formulas, but not too complex for AI. NN can even imitate the phenomenon of synesthesia (i.e. translate the colors to the language of sounds and vice versa).

What can go wrong

It’s hard to imagine what contemporary art should be further into the XXI century, in the era of total technological reproducibility and instant «improvement» by the mega-brains of NN. The abstract art like Kandinsky or Pollock is over, because NN can generate hundreds of them in a moment. Also, it is very possible that a lot of low-tier “art jobs”, like design or video editing will be outsourced to robots.

What can you do

The two things the machines can’t do and never will: to think and to care. And that’s what all the artists or content creators should stick to, to stay relevant in the technological future.

Neural network and doctors

How it works

The importance of telemedicine and the newest technological solution for diagnostics and analysis was made obvious during the COVID-19 outbreak in 2020 Spring. The scientists made a neural network that can diagnose COVID by coughing – the difference is invisible to the human ear, but a special algorithm is able to find these crucial differences. Another NN was taught to diagnose cancer more effectively than human doctors by analyzing the X-ray photos.

Neural Networks in Medicine
Neural Networks in Medicine

Yet another group of scientists in 2020 combined machine learning methods and chemical equations. This provided the opportunity for a more accurate prediction of the characteristics of molecules, which is used in the computer design of new drugs or materials.

And that is the future. Neural diagnostics & neural synthesis to support our human bodies alive and well.

What can go wrong

While NN can help save a lot of effort and time for the doctors, it will never be the real substitute for them. The real danger of neural network technology is that people will rely on it too much, though it’s a matter of a decade or two before NN in medicine will be trusted as much as in banks.

What can you do

Trust technology but always listen to the human specialist when it’s a matter of your own health.

Neural network as decision-makers

How it works

No sane person (or even thousands of outsourced foreign workers) can actually process the big data manually. That’s why it’s called “big”. There are hundreds of thousands of posts in social media that appear every minute and human moderators can’t process them all. Simple censoring algorithms (like keyword censoring) are annoying and easy to cheat. NN will solve this, and no hate speech will be allowed online, ever.

HR? Neural networks can analyze applications better than humans and much faster.

Banking? There are neural networks that, based on data on banking transactions, can assess the creditworthiness of customers. Also, probably better than human bank workers.

What can go wrong

AI makes mistakes. It copies not only the useful data but also all flaws of its creators: in 2018 Amazon closed the NN project that discriminated against female applicants. AI wasn’t guilty: the western HR model was. Since statistically 93% of IT workers in the USA are male, NN decided that being male is a “default” positive trait of the applicant.

What can you do

As wild as it may sound, we probably should start learning to live the way that can be measured by the AI. Neural networks know no ambiguity, only zeroes and ones. It’s still a computer mind, after all. Think as a computer to understand its logic.

Conclusion

We don’t know what impact the use of neural networks will make to humankind. The experiment of transforming everything under the sun into a great big data pool is quite unique. And while it’s probably a dangerous turn – to hand our society and daily lives to cyber brains – it’s the only way to actually learn more about NN’s possibilities, and about the patterns of our own lives.

Author: Nicolas Cuts

Product Managers at SwitcherryVPN. Have 5 years background in management and marketing. I never stop learning!

Leave a Reply

Your email address will not be published. Required fields are marked *