In his remarks at the Senate AI Insight Forum, Mark Zuckerberg framed a partnership between the government and business as advantageous for American interests.
The CEO of Meta, Mark Zuckerberg, spoke at the Senate AI Insight Forum behind closed doors in an effort to support sustained American leadership in creating AI as a secure path to delivering the economic benefits of AI.
He expressed his fervent support for corporate America’s involvement in establishing the technological norms for both open source and closed source AI development.
Mark Zuckerberg especially promoted the benefits of open source AI:
“…open source democratizes access to these tools, which helps level the playing field and foster innovation for individuals and businesses, which I think is valuable for our economy as a whole.”
Zuckerberg Supports Innovation And Safeguards
Innovation and safety were two of the topics the CEO of Meta discussed right away.
He also emphasized that ChatGPT is not the only type of generative AI.
He promoted regulation of all types of AI, including robots, vision, and natural language processing.
His remarks suggested that he welcomed scrutiny and that there are two problems that need to be resolved:
American technological leadership in AI Safety
Open source AI was supported by Zuckerberg as a means of democratizing technology.
Because businesses intentionally include safety alignment into their products and because open source is collaborative, it aids in the identification of undesirable or unsafe consequences, he praised both closed source and open source AI as being safe.
“…businesses must ensure that goods are developed and deployed properly.
At Meta, we incorporate safeguards into our generative AI models and solutions from the outset and engage with others to set up barriers.
We believe that collaboration among governments, academia, public society, and business will help to both avoid the possible risks and maximize the potential advantages of this new technology.
Since more individuals may examine it to spot problems and then share and spread fixes that can be used to harden systems, it is generally agreed that open source software is safer and more secure.
However, recent instances cast doubt on the claim that opensource software is safer and more secure.
Falcon 180B, an open source LLM that has no safety precautions at all yet is so powerful that it can compete with Google’s most potent AI, Palm 2, was just released by Hugging Face.
Falcon 180B was made available by Hugging Face as the base version with the idea that it would be up to the users to make it secure.
Zuckerberg’s American Leadership Agenda
Additionally, Zuckerberg pushed for an American leadership vision for AI.
In terms of AI innovation, America now leads the world.
The amount of human resources used by Silicon Valley businesses in the United States eclipses that of any other nation.
The scope of American creativity is unmatched by any other nation or region.
In order to foster a European-based AI sector and further solidify American leadership, OpenAI recently announced that they are creating an office in Dublin.
However, Zuckerberg supported the notion that the US government and business should collaborate in order to sustain technical innovation and leadership in the country.
As Zuckerberg clarified:
“…America must maintain its position of leadership in this field and set the global technical standard, in my opinion.
Abu Dhabi is developing the next major open source model, and other nations are also working on it.
I think it’s preferable for American businesses to set the example because they can collaborate with our government to develop these models on crucial topics.
When referring to Artificial General Intelligence (AGI), or superintelligent AI, Mark Zuckerberg was oddly evasive about what should occur.
A type of AI known as AGI is capable of learning and reasoning like a human in a very independent manner. It can also solve complex issues in the same way that people do by drawing on existing knowledge.
Even the definition of AGI is up for debate, though Sam Altman defines it as an AI that can develop science in ways that are not conceivable for humans.
“Now, these equities will change and we’ll reconsider this approach if at some point in the future these systems approach the level of superintelligence,” the author said.
In the end, Zuckerberg expressed a positive outlook for partnering with the government.
Although it is too early to suggest what that partnership, oversight or regulation should look like, Zuckerberg’s remarks appeared to welcome that partnership as a way to help the United States to continue holding the technology lead in artificial intelligence.