- Hindi News
- Career
- OpenAI Researcher Whistleblower Suchir Balaji Commits Suicide Behind Story Explained Major Issue With AI
20 hours agoAuthor: Shivendra Gaurav
- copy link
I thought AI was something that could be used to solve difficult problems – like treating diseases and preventing aging. I thought that we could create a scientist who would help in solving these problems. Then when lawsuits started being filed against AI companies I started worrying about it.
This was said by Suchir Balaji, researcher at Artificial Intelligence (AI) firm OpenAI. On November 26, 26-year-old American Indian Suchir died. He was found dead in his apartment.
Suchir’s allegations added to OpenAI lawsuit 1 day before death
This matter came up for discussion on 14 December. Police have not found any evidence of any foul play in the initial investigation. Police suspect that 26-year-old Indo-American Suchir had committed suicide.
Suchir came into limelight for the first time when he expressed concern about the misuse of AI and its protected content. He also talked about violation of ethical rules in OpenAI. While he himself was one of the people who created OpenAI. Suchir had accused OpenAI, the company that made ChatGPT, of violating policies.
Suchir’s name was included in a lawsuit filed against OpenAI on November 26, a day before his body was found. According to media reports, after this OpenAI had agreed to review Suchir’s file in which he had expressed concerns related to copyright.
Suchir Balaji
Suchir wanted to cure diseases and prevent old age through AI.
A startup named DeepMind had created an AI software that could play games like Space Invaders, Breakout and Pong on its own. The news of this startup aroused Suchir’s interest in AI. Suchir wanted to solve people’s problems through AI.
Speaking to the New York Times, he said, ‘I thought we could make AI a scientist who could help cure diseases and prevent ageing.’
At this time Suchir was a computer science student at the University of California, Berkeley. He started the work of developing a mathematical system called Neural Network. He wanted to make it such a network that could learn some skills by analyzing a lot of data.
In simple language, Suchir was trying to create a software in which data could be fed on how to solve different problems. After that the software can learn the skills to solve those problems on its own. In general, all the AI tools we have access to today do the same thing. For example, AI has learned from us to write good stories in English on any topic. AI is still learning to work in Hindi.
Suchir Balaji joins OpenAI, played an important role in creating ChatGPT
Elon Musk along with Sam Altman created OpenAI in 2015. Its head office is in San Francisco, USA. It is an artificial intelligence (AI) research company that creates AI tools and services like ChatGPT.
Sam Altman (right) with Elon Musk (Photo Source – Getty)
Musk resigned from the company’s board in 2018. Musk had also sued OpenAI and several other people in the company, including Altman. He accused the owners of OpenAI of violating contractual agreements.
Suchir leaves OpenAI job in August 2024
Suddenly in August 2024, Suchir decided to leave OpenAI. While leaving the job, he said that the technology he had helped develop was doing more harm than good.
He said,
If you believe the same as I do then you should also leave the company.
In the lawsuit filed by Musk, it was also said that Altman and Greg Brockman had approached Musk to create an open source, non-profit company. This company was going to develop Artificial Intelligence technology for the benefit of humans. However, due to OpenAI’s focus on making profits, the agreement between Musk and OpenAI broke down.
What concerns did Suchir express behind leaving OpenAI?
Suchir had worked with OpenAI for 4 years. While leaving the job, Suchir had expressed three big concerns-
1. Other firms, including OpenAI, have illegally used copyrighted data of computer programmers, artists, writers and news organizations to train their chatbots and AI tools.
2. ChatGPT and other such AI chatbot tools can pose a threat to content creators, businesses, and internet services. Whereas AI took training using the data created by them.
3. AI technologies are replacing existing internet services. Whereas AI tools often provide wrong and fabricated information. The internet is getting worse because of this.
While talking to The New York Times, he told that systems like GPT-4 make a complete copy of the data with which they are trained. Once all the data is copied into the software, companies like OpenAI can train the system to produce either a copy of the same data or an entirely new output.
He said, ‘The reality is that between taking data and then giving output, companies train AI to do something different. The output that AI produces is not an exact copy of the input, it is not even a completely new output.
Since then, Suchir has been continuously talking about the dangers of ChatGPT on his X handle and his blog. In a post on X in October 2024, Suchir said,
Initially I did not know much about copyright, fair use of internet content etc. I became more concerned about this when lawsuits started being filed against AI companies.
Suchir’s X Handle
Suchir’s claims were refuted by OpenAI
OpenAI said in a statement, “We build our AI models legally, using publicly available data, under fair use and regulations. Our philosophy is fair for creators, important for innovators, and America. It is important for competitiveness.
What is ‘fair use’, which Suchir called a false defense of AI companies
Using someone’s content voluntarily or by legally purchasing his rights is called fair use. Suchir had said something very worth considering regarding this fair use. Something was said which is very worrying for content creators, newspapers, news websites, writers and artists. He said,
Saying there is fair use in AI models is like a defense that cannot be trusted. Because the data on which AI is trained, it can use that data to create a product that can compete with that data.
It is such a thing that AI steals AR Rahman’s music and makes music similar to it and even better. Then when AR Rahman strikes copyright on it, it should be said that this is the music created by our AI.
If content creators, newspapers, news websites, writers and artists also apply this example to themselves, then they can understand how justified Suchir’s concern was.
Suchir has exposed the alleged fair use of AI models in his blog.
Also read this news:
Ding’s blunder that made Gukesh the world champion: What is the king-pawn endgame, of which Ding Liren became its victim in the 55th move
12 December 2024. Final match of FIDE chess tournament in Singapore. The score was tied in the last 13 matches between 18-year-old Gukesh of India and defending champion Ding Liren of China. Had this match also been a draw, the world champion would have been selected through speed chase. Read the full news…