OpenAI criticised for lack of transparency around GPT-4

16 Mar 2023

Image: © sofirinaja/Stock.adobe.com

The company is not being open about GPT-4 due to the ‘competitive landscape and safety implications’ of large language models.

OpenAI is facing a wave of criticism online for not disclosing the training details behind its latest AI model, GPT-4.

The company behind ChatGPT caused yet another shockwave in the AI sector this week when it released its latest large language model. The company claimed this model has shown to be more accurate and less prone to “hallucinations”.

But OpenAI’s technical report that came with the release of GPT-4 reveals that the company is keeping various details private, such as the model’s architecture, hardware and training methods.

The company said these details are being hidden due to “both the competitive landscape and the safety implications of large-scale models like GPT-4”.

The market has grown more competitive in recent months, as the sudden popularity of ChatGPT has created an AI-arms race among many Big Tech companies, such as Google revealing its own Bard model to try to rival ChatGPT.

Anthropic, a start-up co-founded by former OpenAI employees, is also making its own generative AI chatbot Claude available to businesses as an alternative to ChatGPT.

But OpenAI’s decision to not reveal details about GPT-4 has caused concern among some in the AI community. Ben Schmidt, the VP of information design at Nomic AI, tweeted “we can call it shut on ‘open’ AI”.

Speaking to The Verge, OpenAI co-founder Ilya Sutskever said the company was “wrong” to previously share its research publicly.

“I fully expect that in a few years it’s going to be completely obvious to everyone that open-sourcing AI is just not wise,” Sutskever said.

OpenAI said it plans to make more technical details available to “additional third parties”, so they can advise on how to “weigh the competitive and safety considerations” against the “scientific value of further transparency”.

Partnership with Stripe

Meanwhile, Stripe has revealed that it is working with OpenAI to help monetise its flagship products, including ChatGPT and Dall-E.

The fintech said it is using its billing and checkout services to power ChatGPT Plus, which is the premium subscription offering of ChatGPT.

Stripe Billing is also being used for DALL-E users, who are able to purchase credits to generate new images.

The fintech is also one of the early adopters of GPT-4 and has already shared how it will enhance Stripe Docs with the AI model.

“Like the introduction of email, smartphones, or videoconferencing, GPT-4 has the potential to fundamentally rewire – and improve – how businesses run,” said Eugene Mann, product lead for applied machine learning at Stripe. “By integrating GPT-4, Stripe is giving our users the most advanced tools to help them build and grow online.”

10 things you need to know direct to your inbox every weekday. Sign up for the Daily Brief, Silicon Republic’s digest of essential sci-tech news.

Leigh Mc Gowran is a journalist with Silicon Republic

editorial@siliconrepublic.com