Generative AI

Generative AI

News: In the wake of newly released models such as Stable Diffusion and ChatGPT, generative AI has become a ‘hot topic’ for technologists, investors, policymakers and society at large. 

What is Generative AI?
• Generative AI is a type of artificial intelligence that involves creating new, original content or data using machine learning algorithms.
• It can be used to generate text, images, music, or other types of media.

What is GPT?
• A Generative Pretrained Transformer (GPT) is a type of large language model (LLM) that uses deep learning to generate human-like text. 
• It is called Generative because they can generate new text based on the input they receive. 
• It is referred “pretrained” because they are trained on a large corpus of text data before being fine-tuned for specific tasks and “transformers” because they use a transformer-based neural network architecture to process input text and generate output text.

How does Generative AI help?
• Language and Content - Generative AI can be used to generate natural language text. For example, researchers are using ChatGPT to generate news articles, poetry, and even code. These models can be fine-tuned to use a specific writing style, or to generate text based on a specific topic or theme.
• Computer Graphics -  Generative AI can be used to create realistic images and animations. For example, researchers have used generative models to create photorealistic images of faces and animals, and to animate virtual characters in real-time.
• Music and Art – It can be used to compose music and create art. For example, researchers have used generative models to create new pieces of music that are similar in style to a given artist.
• Potential Drug Discovery - AI-assisted drug discovery has the potential to accelerate the drug discovery process significantly. Generative AI models can be used to predict the properties of new compounds and predict their potential efficacy as drugs.
• Robotics – It can be used to design and control robotic systems. 

What are issues associated with Generative AI?
• Generative AI has the potential to displace jobs and lead to job loss. For example, generative AI models can be used to automate tasks that were previously done by humans, such as writing news articles or composing music.
• Generative AI can reduce the need for human cognition, as the models can perform tasks that would otherwise require human intelligence. This could potentially lead to decrease in overall cognitive abilities of significant population especially the children who will rely on AI to do their homework. 
• It can also lead to the replication of societal biases present in the mind of its trainer or programmer, this can also have a negative impact on marginalized groups and lead to further discrimination. 
• Fear of Misinformation and Mistrust by manipulation of information, creating fake text, speech, images or video. 
• Fear of Concentration of Power in the hand of a few companies.
• Risks for national security using automated troll bots, with advanced capabilities.

Way Forward:
• Developing ways to make the inner workings of generative AI models more transparent, so that the public can understand how and why the model is making certain decisions. 
• Accountable governance especially of Big-Tech companies using a designated “AI ethicist” or “AI ombudsman”.
• Designing systems that involve human oversight and decision-making at key points in the generative AI process, in order to ensure that the model is behaving as intended and that any unintended consequences can be quickly identified and addressed. 
• Encouraging collaboration between human and AI in the decision making process where AI can be used as a support system while humans make the final decision, which will keep humans in control over the outcome. 
• Continually assessing the potential consequences of generative AI models, both intended and unintended, and making adjustments as needed to minimise any negative impact is necessary.