The probability that AGI will appear 30 years from now is around 80%
– Sam Altman at the OpenAI Hackathon talk
A grey homely building quietly stands in San Francisco’s Mission District. Hundreds of people pass by the building every day, and very few of them, possibly none, know that truly groundbreaking sci-fi work is happening there….for real — the creation of artificial general intelligence (for those skeptics who claim that kind of technology is impossible, I’d remind them of a quote from Lord William Thomson, mathematician and physicist, who stated in 1895, “Heavier-than-air flying machines are impossible.”)
Perhaps, thanks to people who work inside the building, humanity will obtain the most powerful technology that has ever been created and will be able to design its own evolutionary descendants. But with this power comes a high risk of abuse. It can turn against humanity rather than be an advantageous point of leverage. With the mission to make AI safe for humanity, a little over two years ago Elon Musk, Sam Altman, Ilya Sutskever and Greg Brockman launched OpenAI. Since then, the organization has become a key player in the AI scene, gathering together the world’s top AI researchers and engineers.
To inspire the AI community and bring smart people together to work on the mission of building intelligent and progressive AI technology, this past Saturday OpenAI hosted the very first hackathon. More than 100 graduates, professionals and everyone who is contributing to AI technology attended the event.
Sam Altman, Y Combinator President and Board Member of OpenAI, in his opening talk covered the most sensitive topics around AI, including the potential US vs China AI arms race.
“This [AI arms race] is particularly bad. One of the things we are going to do here is to build up enough of a political influence, a coalition of different forces, to make that not the only option.’’
Sam did note, however, that China is absolutely surpassing the US in AI technology development. Along the way he gave an example of a mid-level Chinese minister whose group includes thousands of researchers working for government needs, emphasizing how seriously they are taking the emerging technology.
Indeed, China has revealed a plan to become a leader in AI by 2030, making it a domestic industry worth almost $150 billion. In a centralized decision-making system, a given industry will be fueled with billions of dollars and developed with unheard of speed if there is the political will.
Consider an example: China has launched the first ever world quantum satellite to transmit ultra-secure data, inaugurated a 1,243-mile quantum link between Shanghai and Beijing, and announced a $10 billion quantum computing center.
But at the same time, the centralized approach presents serious drawbacks. And perhaps the next example is not so representative because it happened 500 years ago, but it showcases the vulnerability of a ‘one-person’ decision-making approach. At that time the Chinese elite destroyed (literally burned) its world-dominating navy as its political forces were afraid of free trade. It won’t be surprising, comparatively, if some AI initiatives will be totally annihilated in the future because of the willpower of one political leader.
To conclude, Sam Altman noted that “AGI won’t happen tomorrow, but much quicker than many people realize” and predicted the probability that super-intelligence will appear 30 years from now is around 80%.
To that end, Ilya Sutskever, Co-founder and Research Director at OpenAI, noted in his presentation that the self-play approach (an agent playing against itself as many times as possible) will be one of the most important components of AGI.
At the end of the 8-hour hackathon a variety of projects were presented from AI safety to evaluating the electrical density of molecules using object recognition, to the application of generative models such as VAE for data augmentation, to launching the Wikipedia for Deep Learning.
