Past Performance is Not Indicative of Future Results (2020)

Many people are very resistant to the idea that their particular work can have a negative impact or that they should take responsibility for that. See Yan Lecun quitting Twitter (https://syncedreview.com/2020/06/30/yann-lecun-quits-twitter...)

Other people are very aware of the dangers of their work. But, when the money gets big enough, they take their concerns to the bank and their therapist. See Sam Altman's concerns about the dangers of machine intelligence before he invested in OpenAI (https://blog.samaltman.com/machine-intelligence-part-1) Contrast that with his decision to become the CEO, take the company private and license GPT-3 exclusively to Microsoft. (https://www.technologyreview.com/2020/02/17/844721/ai-openai...) He had reasons. He posts here. He might defend himself. But to me it seems like the kind of moral drift I've seen happen when people in silicon valley have to make hard choices about money and power.

There are also applications of ML that are generally safe and can be of benefit to society. See the many medical uses including cancer detection.(https://www.nature.com/articles/d41586-020-00847-2) Most of the work being done to expose the risks and biases of ML is being done by researchers who are at least somewhat within the field. In my math and computer science program, two and a half of the 25 students are doing their thesis in safe ML. (I'm giving myself a half because I'm working on logic based ML.) I don't think it's fair to believe that every person working in ML is participating in something negative for society.

Ultimately, I think we need some reasonable regulation and a lot more funding for research into safe ML. Corporations and governments want ML for purposes that can be unethical. Unfortunately they also control a lot of the research grants. So they have a disincentive to fund AI ethics or safe ML over pushing the boundaries of what ML can accomplish.

Finally, I think many engineers would like their work to be positive for society. Unfortunately, with what we know now, a lot of the edge cases we run into are unfixable. When Google Photos started classifying black people as gorillas, Google just removed primates from the search terms. Years later, they hadn't fixed it. (https://www.wired.com/story/when-it-comes-to-gorillas-google...) I'm sure most engineers on the project knew that was a hack. When faced with an unfixable issue like that, the engineer either tries to get the company to stop using ML for that problem, compartmentalizes and ignores the issue, or they quit. Where do you draw the ethical line? It's good to hold people accountable but it's unrealistic to expect that to solve the problem.

Three remarks:

- The task many people seem to be benchmarking against is not just a measure of general intelligence, but a measure of how well AI is able to emulate human intelligence. That's not wrong, but I do find it amusing. Emulating any system within another generally requires an order of magnitude higher performance.

- The degree to which human intelligence fails catastrophically in each of our lives, on a continuous basis, is way too quickly forgotten. We have a very selective memory indeed. We have absolutely terrible judgment, are super irrational, and pretty reliably make decisions that are against our own interests, whether it's with regard to tobacco use, avoidance of physical exercise, or refusal of life-saving medications or prophylactics. We avoid spending time learning maths and science because it's not cool, and we openly display pride in our anti-intellectual behaviours and attitudes. We're all incredibly stupid by default.

- AI researchers need to work more closely with neuroanatomists. The main thing preventing AI from behaving like a human is the different macro structure of human NNs vs artificial NNs. Our brains aren't random assortments of randomly connected neurons: there's structure in there that explains our patterns of behaviour, and that is lacking in even the most modern AI. We can't expect AI to be human if we don't give it human structures.

If I read all day everyday from the moment I was born until now, I couldn't have read 45 terabytes of text.

1) Regarding the example of qualitative data via drunk students attending eye-licking parties. The author doesn't explain how this is qualitative. To my mind it's a gap in the model. The modelers could have included parameters to account for students behaving impulsively or irrationally, but they didn't.

2) Considering the nebulous nature of terms like consciousness and comprehension and the ensuing challenges of measurement, can it be proven that the structures that can potentially underpin behavior, etc., that would generally be recognized as conscious or comprehending do not exist as an emergent but undetected property of the Internet? Is it reasonable to suppose that if such a structure existed and if it possessed or embodied consciousness or comprehension that it might work toward remaining unknown?

IMO the main difficulty is that humans have terrible self-awareness or self-insight. We want to believe we're special, we want to believe we're intelligent, we want to believe we're different than machines. We're in denial about that.

Our brains aren't any more special than computers, other than that it's really quite formidable that we evolved them by chance in this universe of chemical soup we find ourselves in. At the end of the day, however, a computer is a computer, and "thinking" and "consciousness" simply do emerge from low-level computations given some special structures.