Artificial Intelligence - Branches and Outcomes

Technology is growing rapidly and gaining prominence in all industry sectors. The implementation of a technology, though slow in progress, over time, can spread and provide growth opportunities  through in-depth testing, proper laws in place, and designing safety practices.
Artificial Intelligence is a branch of technology which has the potential to provide growth and the power to connect and take human growth to the next level. It is the culmination of all branches of learning, which enables fusion of all knowledge to build an agent that not only can provide services, but adapt to provide better services via constant learning.
​The two popular branches in non-stop buzzing AI world are Generative and Agentic AI. Both forms of AI use Machine Learning (ML) and Large Language Models (LLM’s) as input data. While both provide necessary services via text, and images, based on user instructions that data scientists term as “Prompt engineering”. Both Agentic and Generative AI have distinctive features and are designed to provide specific services.

www.youtube.com/watch?v=ysDicZyZdMkGenerative AI is designed to create new content, and images based on a set of data that is already fed into the system. This data is called the training data and Generative AI produces output which are mostly predictions for the user instructions by recognizing patterns. A chatbot is a good example of a Generative AI.
Agentic AI is an Advanced and complex artificial intelligence platform that is designed for specific tasks. An agentic AI designed for tax purposes may not be helpful when a user input question is about coding or math. An Agent AI uses reinforced learning(RL) where it provides answers and it is rewarded or penalized based on its choices. It continuously learns and adapts its answers based on the environment.
​An Agentic AI is like a planner and  Generative AI is like an action taker. Both Generative and Agentic AI can be combined together to offer better services to the customers.
Checkout the video - Agentic and Generative AI

Deepfake
Deepfake in Artificial Intelligence is a technology to create made-up images, videos, and audio content. The content does not exist in reality and it is all a fake.
Deepfake uses machine learning to process data from various sources (such as websites, social media, manual input source) in multiple levels and generates best possible new content. The Deepfake AI model is trained with real audio, video and image content and sent to a neural network, where the real magic happens and new content is generated.
Deepfakes are produced using two different AI deep-learning algorithms. One algorithm creates new content that resembles the real image, video, or audio, and the second algorithm detects if the generated new content closely resembles the original and gives feedback to the first algorithm. Based on the repeated feedback the first algorithm refines the content to make it more real.
There are positive and negative uses of deepfake algorithms. The positive outcomes include spreading awareness about social issues, in education, and in medicine.
Positive Outcome examples:-

1. Soccer player David Beckham created a campaign to spread awareness about Malaria.  A Deepfake video was created  with him speaking in multiple languages. This broadened his message to a wider audience.
2. In classroom, deepfake audios can be used to create historical speeches to give a more immersive experience.
3. In medicine, Deepfake technology can be used to detect where tumors can originate.

The negative outcomes of deepfake technology can include creating and spreading disinformation thus creating confusion among people.
Negative Outcome examples:-

1. Apart from celebrities, politicians, and CEO’s, ordinary citizens can be attacked by creating wrong content about them.
2. Deepfakes can be used to alter existing art and music to create new art and music.

One of the ways to identify deepfakes is through News Literacy. News literacy is thinking like a journalist which involves thinking critically and analyzing what information to trust by fostering healthy skepticism.
Tips to spot Deepfake:-

1. Check for backgrounds, such as lighting, shadows in the image or video that is not quite right.
2. Check for some wrong moves by the person in the videos or around the edges.
3. Check for spotless or unnatural face and gestures.
4. The voice may not match the lip movements
5. AI tools such as Deepware scanner, Sensity AI, and Microsoft video authenticator etc.. are developed to detect deepfakes.
6. Check the source of the video.

Checkout the video on Deepfake - Deepfake video