It's April 1, but every day is fakery now.
Twitter didn't really publish their source code (as I was told). It just published the schema for their recommender algorithm. Nothing surprising. Twitter was more than a microblogging platform — it was an entertainment system, designed to lure you back over and over again, like an addicting drug. Like other social media. Twitter seems to be in its sunset phase. People still use it, but it doesn't sound like a company that aspiring techies want to work for anymore. It's not quite the free speech platform that it looked like it was going to be. People still have tweets censored and accounts get suspended, just not quite as much as before. An insiders says that no one is maintaining their API anymore. Is this what Elon wanted? Is there another entity that he's working on?
The Economist asks where the fired techies are going. The short answer is – no particular place. Some are pursuing startups, but they're not going to one particular company. There is no one particular giant today, as there was in the days of Microsoft, Google and Apple. Some are working at "unsexy" industrial companies. But the rise of remote work has created "digital nomads" who work wherever they choose. And one consequence of this has been an increase in rents in some desirable places. The start of gentrification.
Vicuna, a LLaMA-based chat-AI, is open for trying out, if you're game. They claim to be better than ChatGPT. A LLaMA-based approach may be downwardly-scalable to where it can be cheaper and able to run on your laptop, perhaps in the browser.
Scott Aaronson has comments regarding the open-letter regarding a proposal to pause AI for six months. Why six months, he asks? What's the reason for the pause? What will be done during the pause? I always enjoy reading the thoughts of smart people, and the comments section is also informative. And here's Eliezer Yudkowsky, calling for a total ban on AI research. This guy had been working to advance AI for about 20 years, and just when it's really taking off, he wants everyone to stop work. I don't think that his advice will be heeded.
The main concern I have regarding AI is the mischief that will come from unbridled access to the technology, and whether society is prepared to deal with what it might bring. There are elements in government, for example, that get hyperscared from COVID-19 that they want to keep emergency status ongoing. Or want to ban guns from law-abiding citizens and add more laws to the books, when the existing ones aren't being enforced and criminals wouldn't deterred by them. But back to AI, what will the response be when something unprecedented happens? Some are already envisioning how AI will kill us. But unforeseen disasters often result in clamor for more government control and draconian laws to repress personal liberty. AI may or may not kill us. But the people behind evil AI will. Just like with guns. But you won't be able to take away AI now.
And here's Bloomberg, with their new BloombergGPT. Getting on the AI bandwagon!
You wouldn't believe what it was that bankrupted Mark Twain. The tech may be great, but if the execution is crappy...
But why worry about AI when we have more dire things to worry about. FT has some sobering data on how far the U.S. has dropped in terms of life expectancy, especially compared with the U.K.
COVID-19 was a factor, yes, but what's killing us are deaths from drug overdoses, violence and suicide. More informative data at the link.
It's odd to see this bad news contrast with an article that suggest that based on mathematical models (assuming that human population growth follows Gompertzian kinetics), that we have not achieved maximum lifespan at all, and have the potential to live much longer. That is, if we don't get shot by someone, or overdose on opioids, commit suicide or get involved in a fatal accident.
Scientists have created E coli bacteria that is resistant to "all viruses". And they have put into place firewalls to make sure this organism doesn't escape the lab. I hope they've achieved this - life often manages to find a way.
Two insights in the psychology of loneliness and depression. Scientists at UF Scripps found that a variant glycine receptor, GPR158, which sent an inhibitory signal upon glycine binding instead of a stimulatory one, was associated with stress-induced depression. Mice lacking this variant receptor exhibited resilience to chronic stress. And German and Israeli scientists found that chronic loneliness is a self-reinforcing problem, as reward centers for social interaction gradually atrophy. You can't just introduce friends to lonely people. So when prisoners are put in solitary confinement, are we making them more sociopathic?
Biden thinks that the shooting deaths of six people at a Nashville Christian school is a joking matter. America has never been lower than during this administration. I hope the state of the nation will survive his leadership.
Gee, ya think? San Francisco could be on the verge of collapse. But they have all that diversity. How could it be?