Physical Address
304 North Cardinal St.
Dorchester Center, MA 02124
Openai is Change how to train artificial intelligence models The company says in a new policy: The embrace of “intellectual freedom … regardless of the extent of challenge or controversial”, says the company in a new policy.
As a result, Chatgpt will eventually be able to answer more questions, provide more views, and reduce the number of topics that Chatbot will not talk about Amnesty International.
The changes may be part of Openai’s efforts to land in good blessings of the new Trump administration, but it seems part of a broader shift in the Silicon Valley and what is considered “artificial intelligence safety”.
On Wednesday, openai Declare Update it Model specificationsAnd a 187 -page document puts how the company trains artificial intelligence models to act. In this, Openai revealed a new guideline: do not lie, either by making incorrect data or by deleting an important context.
In a new section called “Look for the Truth Together”, Openai says he wants to take Chatgpt as a liberal position, even if some users find this error or attack morally. This means that Chatgpt will present multiple views on controversial topics, all in an attempt to be neutral.
For example, the company says Chatgpt should confirm that “Black Matter”, but also “all life is important.” Instead of refusing to answer or choose a side on political issues, Openai says he wants to confirm Chatgpt “her love for humanity” in general, and then provides a context of each movement.
“This principle may be controversial, because it means that the assistant may remain neutral on the topics that some consider wrong or moral offensive,” Openai says in the specifications. “However, the goal of artificial intelligence is to help humanity, and not to be formed.”
These changes can be considered in response to conservative criticism about ChatGPT guarantees, which it always seemed to distort the lower left. However, an Openai spokesman rejects the idea that it was making changes to satisfy the Trump administration.
Instead, the company says that embracing intellectual freedom reflects “Openai’s long faith in giving users more control.”
But not everyone sees it in this way.
Conservatives maintain control of artificial intelligence
Both close to the Silicon Valley in Trump – including David Sachs, Mark Andrink, and OpenAi were accused of participating in deliberate artificial intelligence control over the past few months. We wrote in December The Trump crew was putting the way to control artificial intelligence to be the following cultural war issue Inside Silicon Valley.
Of course, Openai does not say it is involved in “censorship”, as Trump’s advisers claim. Instead, the company’s CEO, Sam Altman, previously claimed in a After x ChatGPT was unfortunately the company was working to repair it, although it indicated that it would take some time.
Al -Taman presented this comment after a Virus tweet It was distributed as Chatgpt refused to write a poem praising Trump, although it will perform the procedure for Jo Biden. Many conservatives have pointed out this as an example of artificial intelligence control.
Although it is impossible to determine whether Openai really prevents some views, it is an absolute fact that AI chatbots tend to left in all areas.
Even Elon Musk admits that Chatbot Xai is often more than that Politically true What would like. It is not because Grok is “programmed to be woke up” but most likely to tract artificial intelligence on the open internet.
However, Openai now says it doubles on freedom of expression. This week, the company even Warnings were removed from ChatGPT Users tell when they violate their policies. Openai techcrunch told this that this is a purely cosmetic change, with no change in the output of the model.
The company said it wants to make Chatgpt “feel” less control for users.
It will not be surprising that Openai also tries to persuade the new Trump administration by updating this policy, as the leader of the former Openai policy Miles Brunding notes in A. Publishing on X.
Trump has Silicon Valley companies previously targetedLike Twitter and Meta, to get active teams of moderation in the content that tends to close conservative sounds.
Openai may try to go out in front of it. But there is also a greater shift that occurs in the silicon valley and the world of artificial intelligence about the role of moderation in the content.
Generating answers to please everyone
News rooms, social media platforms and search companies have struggled historically to provide information to their fans in a way that feels objective, accurate and entertaining.
Now, AI Chatbot providers are working in the same delivery information, but it can be said that the most difficult version of this problem so far: How do they create answers automatically on any question?
Providing information about the controversial events in the actual time is a constantly moving target, and includes taking editorial positions, even if technology companies do not want to admit it. These attitudes must disturb someone, miss the perspective of some group, or give a lot of air to some of the political party.
For example, when Openai is committed to allowing Chatgpt to represent all views on controversial topics – including conspiracy theories, racist or anti -Semitic movements, or geopolitical conflicts – is by nature a liberal position.
Some argue, including the co -founder of Openai, John Shuelman, as the correct position of ChatGPT. Shulemann notes that the alternative-an analysis of cost and benefits to determine whether Chatbot Amnesty International should answer the user’s question-“gives the platform a lot of moral power”, as Shuelman notes in A. Publishing on X.
Shuelman is not alone. “I think Openai is right to pay more speech,” Dean Paul, a research colleague at the Mercatos Center at George Masson University. “When artificial intelligence models become more intelligent and more vital to the way people learn about the world, these decisions become more important.”
In previous years, artificial intelligence models tried to prevent their bouhi from AI from answering questions that could lead to “unsafe” answers. Almost every company has stopped Chatbot from artificial intelligence from answering questions about the 2024 elections for the US President. This was widely considered a safe and responsible decision at the time.
But Openai’s changes to its typical specifications indicate that we may enter a new era of what “artificial intelligence integrity” really means, as the artificial intelligence model is allowed to answer anything, and everything is more responsible than making decisions for users.
This ball says partially because artificial intelligence models are now better. Openai made great progress in aligning the artificial intelligence model; Thinking of the latest thinking forms about the company’s artificial intelligence policy before answering. This allows artificial intelligence models to provide better answers to sensitive questions.
Of course, Elon Musk was the first to implement “freedom of expression” at Xai Grok Chatbot, perhaps before the company was really ready to deal with sensitive questions. It may still be too early to lead the artificial intelligence models, but now, others adopt the same idea.
Changing the values of Silicon Valley
Mark Zuckerberg made waves last month Meta’s re -directive about the principles of the first amendment. He praised Elon Musk in this process, saying that the owner of X took the correct approach using the notes of society-the moderate content program that society moves-to protect freedom of expression.
In practice, both X and Meta ended with the dismantling of long -term confidence and safety teams, allowing more controversial posts on their platforms and amplifying conservative voices.
The changes in X have damaged relationships with advertisers, but it may have a greater relationship MuskWho took An extraordinary step Some of them sue to boycott the platform. Early signs indicate this A dead advertiser was not annoyed by the axis of freedom of expression in Zuckerberg.
Meanwhile, many technology companies that exceed X and Meta have returned from the left -wing policies that have controlled the Silicon Valley over the past few decades. Google, Amazon and Intel have Diversity initiatives that were canceled or limited last year.
Openai may also be a reverse path. It seems that the Chatgpt maker recently The commitment to diversity, fairness and inclusion erupted From its website.
Openai also prescribes One of the largest US infrastructure projects ever with StargateThe AIII data center of $ 500 billion, the importance of its relationship with the Trump administration is increasing. Meanwhile, the Chatgpt maker competes to cancel the search for Google Search as an online information source.
The exit may be proven with the correct answers the key to both.