Home
News
- 1 GenAI, or Generative AI produces unique content like text, codes, images.
- 2 Language models, transformers and the easy input have given these GenAI immense success.
- 3 Biasness, deepfakes, racial statements, and lack of source are challenges yet to be overcome.
“Hi Harry, this is Linda. Need a formal letter to schedule a meeting with the governor, addressed to his office, ASAP.”
Harry, munching his burger and giggling on the jokes of his colleagues, takes out his phone, hearing a ding, and reads the notification.
Now, an oldish Harry would set aside his recent plans and write the requested letter, on a weekend, cursing his job and boss.
But, a smart Harry would simply open ChatGPT, enter the details, and send the output letter to Linda, and continue enjoying his weekend.
This is a scenario where Harry utilizes the potential of GenAI and you may have used it too on many occasions, either consciously through ChatGPT, Dall-E, Bard, Jasper, Codex,Listnr, or unconsciously when you answer a chat pop up while browsing websites.
Thus, we all admit that we were surrounded by AI (Alexa, Siri) and now a high wave of GenAI is taking us forward.
GenAI: Answering the What’s and How’s
The What:
Generative AI is a content producing AI that can generate texts, audio, images, code.
If you are familiar with ChatGPT, you will know that it first asks for a prompt. This prompt/ input can be a text (Harry’s was to write a letter), an image, some musical notes and then through rigorous machine learning, transformers (one type of neural network architecture) and language models, an output will be generated.
The terms transformer and language model require some description to better understand how GenAI works.
The How:
Consider the sentence, Taj Mahal is in…….
You and I will easily complete the sentence as Taj Mahal is in Agra.
For the AI to complete it, it needs a large database and language training. The database could provide AI, with many images of Taj Mahal and distinguish it. Based on its training, the answer could differ as simply; India or U.P ,India or Agra, U.P, India.
The language model guesses the word that will come next in a sequence of words. This becomes necessary to write sentences grammatically correct and in a tone sounding; natural, pleasant and human written.
The Enormous Popularity: Reasons and Discussion
ChatGPT is by far the most popular GenAI presently, but it’s not the earliest and the only of its kind.
Eliza, a chatbot developed in the early 1960’s by Joseph Weizenbaum, takes credit for the first GenAI model.
The recent success of models now, can be attributed to the ease in input methods they have brought. Previously, the input was through an API or some complex methods, but now it’s as simple as chatting with a person.
Some popular GenAI, extending their services presently are:
ChatGPT: Created by OpenAI, and launched in Nov ’22, this tool took the world by storm mainly due to its simple UI and enormous database.
It uses a chat interface and can customize the results based on user input.
Dall-E: Its an image creating tool, which creates unique images based on user input. It eats its share of pie in its capability to convert words (the user input) into image (output).
BARD: How can Google be behind in this domain, and it has launched its GenAI tool, BARD. But it did not gain immense popularity because of some inaccurate results.
Future, Challenges and Ethics
The answers provided by GenAI lack a source, and as a user, we often ask ourselves can it be trusted? Cases of misleading and false information are already piling up on some GenAI which worries the user.
The text created can be biased based on gender/religion, may contain racial statements which upon reading by young students can plant a seed of the above said. Deepfakes are already circulating on social media, which furthers our concern over these GenAI.
Nancy J. Allen is a crypto enthusiast, with a major in macroeconomics and minor in business statistics. She believes that cryptocurrencies inspire people to be their own banks, and step aside from traditional monetary exchange systems. She is also intrigued by blockchain technology and its functioning. She frequently researches, and posts content on the top altcoins, their theoretical working principles and technical price predictions.
