Transparency is crucial to how AI is trained – regulators must lead tech news

Transparency in developing AI systems is critical, but efforts to improve AI systems must be led by regulators, not private companies.

Nick Clegg, head of global affairs at Meta, today sees openness as the way forward, telling the Financial Times that more transparency on how issues are being dealt with artificial intelligence The work “is the best antidote to the fears surrounding technology”.

Since its launch in November last year, Chat GPT Captivating the imagination of the public with its ability to quickly answer user questions in a personable manner.

The app is an example of generative artificial intelligence, which can generate text or other media based on prompts.

It was trained by OpenAI in September 2021 on a large collection of internet texts, books, articles and websites.

The problem is that the company doesn’t share information on how the chatbot was trained, so it can’t directly fact-check its responses.

Its peer Meta sees the recent decision to make publicly available 22 “system cards” that offer insight into the AI ​​behind ranking content on Facebook and Instagram a step toward greater transparency.

More about artificial intelligence

However, the system card itself only provides a surface view of how Meta’s artificial intelligence system will be used.

They do not have a comprehensive understanding of the level of responsibility for the process of designing these systems.

Please use Chrome browser for a more convenient video player

How Sky News created an AI reporter

David Leslie, research director for ethics and responsible innovation at the Alan Turing Institute, the UK’s National Institute for Artificial Intelligence, said the cards provide a “bird’s eye view”.

“It will talk about how the data is collected, it provides very general information about the components of the system and how some choices are made,” he said.

read more:
Martin Lewis warns of ‘terrible’ AI scam videos
Microsoft boss says AI ‘is not equipped to take over’
How artificial intelligence is changing the future of journalism

Some might see this as a first step, but in an industry where controlling access to information is an essential source of business revenue, companies have little incentive to divulge trade secrets, even if it is necessary to build public trust.

Please use Chrome browser for a more convenient video player

AI can now speak human language fluently

So far, there is no policy regime in place to force enough transparency about AI from private companies.

However, following calls from campaigners, the UK is gearing up for a second reading in the House of Commons in November with a private MPs’ bill.

The next step for regulators is to provide specific guidelines on what information can be accessed and to whom, to enhance accountability and protect the public.

Source link