Why is No One Taking Generative AI Seriously?

Although there are plausible criticisms of generative models, a lot of models like ChatGPT, DALL-E, and Stable Diffusion do have real-world applications and proved to be useful in many scenarios
Listen to this story

Is generative AI – particularly ChatGPT or DALL-E – just too much fun to mean business? The fun part is attained through real-time content creation. So is it entertaining? Yes! Useful? Not so much – resonates Meta’s chief AI scientist Yann LeCun. 

In an exclusive interaction with Analytics India Magazine, LeCun said that these systems in their current form are just for entertainment, and do not lead to anything useful. Further, he said that for generative AI to be useful, it has to make sense of real world problems and assist people in their daily lives. “I don’t think these systems in their current state can be fixed or called intelligent in ways that we want and expect them to be,” said LeCun. 

OpenAI chief Sam Altman also recently tweeted saying that ChatGPT is incredibly limited and creates an illusion of greatness and can be misleading. Though it can be used for fun and creative purposes, relying on it for factual information is not such a good idea. 

Last month, Meta AI and Papers with Code released Galactica, an open-source large language model with 120 billion parameters. Intended for scientific papers and curating a corpus of human’s scientific knowledge, it took only three days for Meta AI to take it down after it started producing unreliable and hallucinatory outputs. Many researchers pointed out that this could be highly dangerous for scientific research.

The long time critic of deep learning for AGI, Gary Marcus, called out Galactica’s “bullshit” outputs and said that Meta AI is following the footsteps of OpenAI’s GPT-3 text generator, which according to him “spits out total hogwash”. 

Read: Top 10 Alternatives to GPT-3

During the Web Summit in Lisbon, Marcus was joined by Noam Chomsky to talk about today’s cutting edge AI. While Marcus was clear and straight up criticised OpenAI’s DALL.E that cannot understand simple grammatical concepts and generates compositionally lacking images, Chomsky was also only a little optimistic about the usefulness of large language models.

Real-World Applications 

Although there are plausible criticisms of generative models, a lot of models like ChatGPT, DALL-E, and Stable Diffusion have real-world applications and proved to be useful in many scenarios. For example, ChatGPT can be very helpful for progressing the edtech industry by providing solutions to simple problems and acting as tutors. 

Some developers tried ChatGPT to generate SQL queries from text and the outputs were perfect. Another scenario was when the model could convert unstructured data into structured data.

Read: These 8 Potential Use Cases of ChatGPT will Blow Your Mind!

Text-to-image generators, though were initially thought to be a threat, are actually benefiting the stock images industry. Recently, Shutterstock announced that it will allow AI-generated images to be used on their website. 

Notion, Jasper, and have been using OpenAI’s GPT-3 and assisting writers to generate comprehensive articles and texts. Recently, Canva also integrated a text-to-image generator using Stable Diffusion. For Cypher 2022, Analytics India Magazine’s flagship AI conference, the banners and posters were also designed using Midjourney.

Apart from text and images, GitHub Copilot has been a blessing in disguise for the developer ecosystem, where it helps in generating code from text inputs. Meta came up with InCoder to match GitHub’s code generator. 

Moreover, StabilityAI announced a collaboration with AWS to make its tools open-sourced for more students and researchers, similar to what OpenAI did with Microsoft. 

Clearly, from a use case perspective, OpenAI seems to be dominating the generative AI landscape. 

Is OpenAI winning the Generative AI race? 

“I don’t think any company out there is significantly ahead of the others,” said Yann LeCun.  He explains that many researchers are working on large language models with just slightly different approaches and that there are three to four companies producing GPT-X-like models. “But, they [OpenAI] have been able to build and deploy their systems in such a way that they have a data flywheel.” 

Download our Mobile App

Mohit Pandey
Mohit dives deep into the AI world to bring out information in simple, explainable, and sometimes funny words. He also holds a keen interest in photography, filmmaking, and the gaming industry.

Subscribe to our newsletter

Join our editors every weekday evening as they steer you through the most significant news of the day.
Your newsletter subscriptions are subject to AIM Privacy Policy and Terms and Conditions.

Our Upcoming Events

15th June | Bangalore

Future Ready | Lead the AI Era Summit

15th June | Online

Building LLM powered applications using LangChain

17th June | Online

Mastering LangChain: A Hands-on Workshop for Building Generative AI Applications

20th June | Bangalore

Women in Data Science (WiDS) by Intuit India

Jun 23, 2023 | Bangalore

MachineCon 2023 India

26th June | Online

Accelerating inference for every workload with TensorRT

MachineCon 2023 USA

Jul 21, 2023 | New York

Cypher 2023

Oct 11-13, 2023 | Bangalore

3 Ways to Join our Community

Telegram group

Discover special offers, top stories, upcoming events, and more.

Discord Server

Stay Connected with a larger ecosystem of data science and ML Professionals

Subscribe to our Daily newsletter

Get our daily awesome stories & videos in your inbox