What is Generative AI? An interesting but ‘scary’ form of artificial intelligence

In the era of modern technology, where artificial intelligence (AI) is talked about, its increasing development is also worrying some people, but despite all the concerns.

People around the world are not investing billions of dollars in artificial intelligence. Don’t hesitate. This?is why Sam Altman, CEO and co-founder of OpenAI, has said?that he?is not concerned about the cost of developing generative AI.

Sam Altman is among those who have played an?important?role in bringing artificial intelligence to the public. After the invention of computers and then the Internet, artificial intelligence is considered the third?greatest?human advancement in?the field of?technology.

Sam Altman was giving a special lecture for emerging entrepreneurs at Stanford University he founded ‘Open AI’ as a non-profit organization in 2015.?

However, OpenAI now says that the cost of technological development cannot be covered by non-profit sources.

As one of the founders of this technology, what he has to say is quite significant. It shows how talented and?powerful?his future types will be.

During his talk, he said that ‘no matter how much investment?is made?in artificial intelligence, it is less?compared to the benefits that the society will get due to it.

What is Generative AI, This type of new artificial intelligence

Generative AI is?a type of?artificial intelligence that can create new?pure content.?This content can be in?any form of?text, images, video, sound or software at the user’s request.

Generative AI relies on highly sophisticated deep learning models of machine learning. Deep learning mimics the human brain in learning and decision-making.

This model identifies desired patterns from vast amounts of data, understands them, and then uses its understanding to create new and?useful?content based on specific user instructions.

Earlier, to give any instructions to the?computer?one had to master different types of computing languages like C++, Java, Python?etc. Now?you?can ask the AI to write code in?normal?spoken language.

Most generative artificial intelligence has three stages

1. Training

In which a base model is built

2. Tuning

A basic model that is adjusted to serve a particular purpose Creation,

3. Evaluation and feedback

In this phase, AI-generated content is continuously evaluated and?continuously?improved for quality and accuracy.

Just as the human learning process never stops, artificial intelligence also has no limit to learning as mentioned earlier, generative AI?is capable of creating?many types of original content.

Let’s see some examples.

Writing

Instructions, product promotions, advertisements, emails, website content, articles, blogs, research papers, and?even?creative writing like poetry. It can provide a summary of a large document.?Can?tell it in points. Thus?it?gives the writer more time to do more creative work.

Images and Videos

Image generation engines like Dill-E, Midjourney and Stable Diffusion can generate any?image?and video you want. You can tell a scene by asking it to be painted by a particular artist or?in a particular?style. You can edit the pre-made video as you want.

Voice and Music

Generative models can combine natural sounds and audio content to speak.?can?read the book. It can produce?completely?original music for musicians, just like a master composer’s tune.

Software Code?

A computer can create code. After?which,?the software developers can quickly remove the errors and produce any software in working condition. To do this, they don’t need to give instructions in computer language but can tell the AI model in human language what code to write and what?it will be used?for.

Design and Art

Generative AI can create unique artistic designs for designers.?Creating?environments and characters for video games?etc.

Illusion and Artificial Data

Generative AI models can?be trained?to generate artificial data and structures based on?real?data.?For example, to develop a new drug, what should its chemical structure be?

The Potential and Threats of Artificial Intelligence

However, as Sam Altman said in response to a question, it should not be considered as good as any other technology. He said that?like?any other technology, it can also?be misused. For example, ‘You can do great things with a hammer, but you can also kill people with the same hammer.’

Generative artificial intelligence is helping scientists find solutions to many unsolved scientific puzzles. Cases of its misuse are receiving far more attention than that It is?being used?to create fake videos, most of which are pornographic.

Fake videos of politicians and heads of state are?being made. It is?being used?for forgery, automated cyber-attacks, and?fraud.

General Artificial Intelligence vs Generative Artificial Intelligence

In recent times, artificial intelligence has taken everyone by surprise. Not long ago, when people talked about artificial intelligence, they meant artificial intelligence that could learn.

It can understand the data?provided,?and make predictions based on it. These data were in the form of examples, sometimes several hundred thousand examples.

Simply put, artificial intelligence can?be taught?to look at hundreds of thousands of X-rays?and?then based on those, it can tell you whether a patient will have cancer?or not.

Now?if?we talk about generative artificial intelligence, it is a step forward.

Now?it?doesn’t just stop at making predictions, it?can even generate an X-ray based on its database to show what an X-ray of a tumour would look like.

How does artificial intelligence learn?

If you’ve used Alexa, Siri, or any other voice recognition system, you’re using AI in some form. You may have noticed how your phone categorizes your photos. It sorts photos by faces.?All this is done by AI.

It looks for patterns in photos and calculates whether the?photo?is of a new person or what is?special?about?the place?where the photo was taken, such as a beach.

The process of machine learning is called training. It involves giving a particular type of data?to a particular computer software. With this data, it?is given?instructions.

For example, it can?be asked?to find images containing faces out of millions of given images.?This?will tell it how to recognize the face. The software will then look for patterns in the?given?data.

It will show what it sees as faces. It may also need to?be corrected during this time It can?be said?that there is no face with such patterns.

The instructions and data given to it become the AI model.?This?determines how effective the AI model will be.

Written By: Arshad Hussain