The Difficult Path to AI’s Promise

This story very much reminds me of Clearview AI that claimed its facial recognition app was "available only for law enforcement agencies and select security professionals to use as an investigative tool" but whose investors had access and used it for their personal whims: Read the NYT article here: https://www.nytimes.com/2020/03/05/technology/clearview-investors.html Even when technology is created with the best intentions, it can be used in ways that run counter to the venture's original mission, and there should absolutely be some closely guarded secrets when venturing into this kind of territory. But who's to say the leadership of OpenAI should have the power to hold these secrets? What oversight is there to ensure ethical use of the technology they've created?

1) Hummm… OpenAI’s “primary fiduciary duty is to humanity.” The company wants humans continue to live meaningful live and it plans on “distributing the benefits” of AGI to “all of humanity,” Who is saying that? Who wrote OpenAI’s charter? Greg Brockman, who had run technology for the payments company Stripe. What strikes me is that the leadership of the organization does not match my expectation of who should be leading a social enterprise with such a noble vision. I just know that Brockman grew up on a hobby farm, went to Harvard and MIT. But were are the artefacts from his life showing a deep commitment to social justice? I remember reading the biography of Leila Jana, CEO of Samasource, another social enterprise. She explained that she grew up poor, saw her parents constantly struggle, had a transformative experience while teaching blind students in Uganda, she studied African development etc…overall there was a clear sense that this person was genuinely committed to work for social justice. But with OpenAi, I think some kind of credible leadership might be missing. 2) From my perspective, their mission is flawed from the outset: “Researchers will be strongly encouraged to publish their work, whether as papers, blog posts, or code, and our patents (if any) will be shared with the world.” The thing is, if they share everything, the “bad guys” (those with somber motives) can build upon OpenAi work to accelerate their own dark agenda.

Hi David. Thank you for sharing this article. A few points really stood out for me. What will Artificial General Intelligence look like? While it is incredibly tough to visualize it, it is safe to say that no matter the technology, it will certainly have its pros and cons. This is evident by the GPT-2 technology that Open AI created. With this technology how does one differentiate between fact and fiction? You make a good point when you say that if we do not understand people, how can we understand intelligence? While Open AI's intentions are commendable, it leaves one to wonder how they can build 'safe systems' that reflect human values and ethics, as these are so subjective and differ from person to person? Another important point raised in the article was on diversity. As Open AI's Research Director emphasized diversity in the work team to be able to "catch everything", is it really possible to do so? Open AI's charter focuses on transparency and not concentrating power in the hands of a few; for all pragmatic reasons it does need money to function and to continue with its research. In my opinion, the company is treading a fine line between being commercial and a social / non-profit venture. While it remains non-profit in spirit, it is forced to confront the fact that money fuels all research. The one thing that I did not understand was why Open AI needs to be the fastest or the first one in the race to create meaningful AGI?