AI learns to write crappy text, and I fail to be alarmed (yet)

One wonders how much of this is media-friendly hype versus actual scary breakthroughs in artificial intelligence:

The creators of a revolutionary AI system that can write news stories and works of fiction – dubbed “deepfakes for text” – have taken the unusual step of not releasing their research publicly, for fear of potential misuse.

OpenAI, an nonprofit research company backed by Elon Musk, Reid Hoffman, Sam Altman, and others, says its new AI model, called GPT2 is so good and the risk of malicious use so high that it is breaking from its normal practice of releasing the full research to the public in order to allow more time to discuss the ramifications of the technological breakthrough.

At its core, GPT2 is a text generator. The AI system is fed text, anything from a few words to a whole page, and asked to write the next few sentences based on its predictions of what should come next. The system is pushing the boundaries of what was thought possible, both in terms of the quality of the output, and the wide variety of potential uses.

Here’s what the program wrote after being fed the first line of 1984:

 “I was in my car on my way to a new job in Seattle. I put the gas in, put the key in, and then I let it run. I just imagined what the day would be like. A hundred years from now. In 2045, I was a teacher in some school in a poor part of rural China. I started with Chinese history and history of science.”

Ugh.

Reminds me of the alarmist and mostly fake stories about how Facebook panicked and “shut down” an experiment in which two bots started to talk to each other in an incomprehensible language. It wasn’t exactly Skynet becoming self-aware, but the headlines tended to make you think that.

Granted, I’m probably not doing the OpenAI research justice here. I’m just very skeptical of AI-related stories that contain words like “fear” and “dangerous” in the headline and lead.

SEC catches up to Elon Musk

On July 27, I wrote that Elon Musk’s “increasingly bizarre and out-of-control behavior of late certainly raises doubts about his qualities as a business leader. The outlook for Tesla does not look good either.”

That was before Musk pulled a little stunt on Twitter that caught the attention of the SEC, and now the tycoon is in serious trouble:

The Securities and Exchange Commission sued Tesla’s CEO on Thursday for making “false and misleading” statements to investors. It’s asking a federal judge to prevent Musk from serving as an officer or a director of a public company, among other penalties.

The complaint hinges on a tweet Musk sent on August 7 about taking Tesla private.

“Am considering taking Tesla private at $420,” Musk said. “Funding secured.”

The SEC said he had not actually secured the funding.

“In truth and in fact, Musk had not even discussed, much less confirmed, key deal terms, including price, with any potential funding source,” the SEC said in its complaint.

That tweet, and subsequent tweets from Musk over the next three hours, caused “significant confusion and disruption in the market for Tesla’s stock,” as well as harm to investors, the SEC said. On the day of Musk’s tweet, Tesla’s stock shot up nearly 9%. It has declined substantially since then.

Tesla’s (TSLA) stock dropped more than 11% in after-hours trading Thursday.

The best part:

The complaint alleged that Musk rounded up the go-private price to $420 per share “because he had recently learned about the number’s significance in marijuana culture” and thought his girlfriend would find it funny. He was dating the musician Grimes.

Just LOL.

Musk update

The Wall Street Journal claims that, according to sources, Elon Musk flipped out when safety sensors automatically halted the assembly line at the Tesla factory in California, and “began head-butting the front end of a car” to prove that the safety measures were unnecessary. He then apparently fired a senior engineering manager involved with the system.

The takeaway here is that Musk is a nutjob, perhaps in the throes of an Ambien-fueled psychotic episode. What sane individual head-butts cars? But the verb choice may be misleading. As the Journal notes later on in the piece: “Tesla said Mr. Musk, in a safety hat, had tapped, not headbutted, the car on the assembly line that day, and that the system was adjusted without jeopardizing safety.” Of course, “Musk tapped the car on the assembly line with his helmeted head” would make for a much weaker hook.

Journalistic hyperbole aside, it does seem that Musk is in over his head (heh), and while I can’t claim to understand what is going on at Tesla, it does not seem good:

The executive team at Tesla that Mr. Musk once relied upon for information is depleted. More than 50 vice presidents or higher have left the company in the past two years.

Mr. Musk hasn’t so far found a second-in-command with the expertise or vision that appeals to him, people familiar with his thinking said.

Mr. Musk had long told his executives he didn’t want to be CEO and planned to serve only as long as it took to bring Tesla up to speed, leaving him to focus on product development, people familiar with his comments said. Aides debated who would take his place: Doug Field, the engineering chief, or Jon McNeill, the sales chief.

Both men left Tesla this year, and Mr. Musk assumed their roles instead of hiring replacements. Tesla said neither man was being groomed for CEO.

“These people are complete narcissists”

Google leadership seminar

Google leadership seminar (source)

I enjoyed this rant against Big Tech, which besides being funny, also contains the kernel of a very interesting idea for how to address the growing crisis around data privacy and ownership:

Bannon also added this gem about Tesla:

I do not have a dog in this fight, but Musk seems increasingly unhinged to me, and the little stunt he pulled with his abandoned buyout plan was undeniably shady. But… are you not entertained?

Daily links: Musk, Mission Impossible, US military

An amusing takedown of Elon Musk. For some reason, Musk is a deeply polarizing figure, viewed as either a visionary genius or a total charlatan. His increasingly bizarre and out-of-control behavior of late certainly raises doubts about his qualities as a business leader. The outlook for Tesla does not look good either.

New Yorker review of Mission: Impossible — Fallout. Very entertaining movie, although the crazily violent fight scenes and endless car/bike chases through Paris get numbing after a while.

All your base are belong to us: More than 300,000 American military personnel are deployed or forward stationed in 177 countries.

More US embassy weirdness: Bomb detonated near the embassy in Beijing.

Some salient questions about the US-EU announcement on trade relations.

North Korea returns remains of (allegedly) US soldiers in goodwill gesture.

What happens when a total stranger decides to destroy your life by posting false information about you on a sleazy grudge-settling website?

Alone in the galaxy

The universe is an even bigger waste of space than we had imagined. Some scholars at Oxford reconsider the famous Drake Equation — an attempt at estimating the number of intelligent civilizations in our galaxy — and find that the number of civilizations in the Milky Way is likely no more than one (emphasis mine):

When we take account of realistic uncertainty, replacing point estimates by probability distributions that reflect current scientific understanding, we find no reason to be highly confident that the galaxy (or observable universe) contains other civilizations, and thus no longer find our observations in conflict with our prior probabilities. We found qualitatively similar results through two different methods: using the authors’ assessments of current scientific knowledge bearing on key parameters, and using the divergent estimates of these parameters in the astrobiology literature as a proxy for current scientific uncertainty.

When we update this prior in light of the Fermi observation, we find a substantial probability that we are alone in our galaxy, and perhaps even in our observable universe (53%–99.6% and 39%–85% respectively). ’Where are they?’ — probably extremely far away, and quite possibly beyond the cosmological horizon and forever unreachable.

Thus, the study suggests that we shouldn’t be terribly surprised if we fail to detect any signs of intelligent life in the universe. In a sense, this is reassuring. If the Drake Equation amply explains the cosmic silence, then we can worry a bit less about the hypothesis of a universal Exterminator that wipes out all life forms that get too smart for their own good.

On the other hand, the strong possibility that we are literally alone in the observable universe makes it all the more necessary for humanity to start pushing out into other neighborhoods. So says Musk:

Elon Musk

@elonmusk
· 25 Jun
This is why we must preserve the light of consciousness by becoming a spacefaring civilization & extending life to other planets https://twitter.com/worldandscience/status/1011081640158945280 …

Elon Musk

@elonmusk
It is unknown whether we are the only civilization currently alive in the observable universe, but any chance that we are is added impetus for extending life beyond Earth
10:15 AM – Jun 25, 2018

The USG will likely be too preoccupied with its own problems over the rest of the century to even think about interplanetary colonization. Can the private sector handle it? Or will the US pass the torch of space exploration to its global competitors?