Daily Tech News 22 January 2206

• Anthropic - the company behind AI tool Claude - has updated its Claude Constitution, now a sprawling 57 page mess, five times the length of the actual Constitution, which has, for the most part successfully, guided the most powerful nation in history for nearly 250 years. (The Register)
  
  Fortunately we are provided with a summary:

  Anthropic hopes that Claude's output will reflect the content of the constitution by being:
  
  

  1. Broadly safe: not undermining appropriate human mechanisms to oversee AI during the current phase of development;
  2. Broadly ethical: being honest, acting according to good values, and avoiding actions that are inappropriate, dangerous, or harmful;
  3. Compliant with Anthropic's guidelines: acting in accordance with more specific guidelines from Anthropic where relevant;
  4. Genuinely helpful: benefiting the operators and users they interact with.

  If Claude is conflicted, Anthropic wants the model to "generally prioritize these properties in the order in which they are listed."

  If only someone had given thought to a prioritised list of general rules - laws, you might call them - to govern the behaviour of autonomous thinking machines - robots, essentially - and all the myriad ways in which things might go awry.
  
  Maybe a Boston biochemist in 1942, while we're engaged in idle speculation.

• Nvidia is expected to launch its consumer-oriented N1X processor this quarter. (Tom's Hardware)
  
  Nvidia hasn't said anything about this, but eagle-eyed observers have spotted the chip in shipping manifests for test hardware, most recently in a Dell laptop.
  
  
  
• OpenAI has committed to "paying its own way", admitting essentially that it hadn't been. (Tom's Hardware)
  
  Specifically for electricity. Everyone else - like people who want to own computers - can fuck off and die I guess.
  
  
  
• The creator of Ruby on Rails says that AI can not yet replace junior programmers. (Final Round)
  
  He notes that sometimes - sometimes - an AI tool can come up with an elegant solution to a small problem, maybe even a better one than you could have built yourself. But overall, 95% of the code his company creates is still built by humans, because at the end of the day for a system to be reliable, people have to understand it.
  
  Which is a complete and welcome turnaround from the demented slopfest of Gas Town which has one simple rule: We don't know what is going on and we don't care. Oh, and a second implied rule: The solution to too much slop is more slop.
  
  
  
• Apple is reportedly working on an AI powered wearable pin. (9to5Mac)
  
  You know that literally nobody is going to buy this thing, right? Even the Tame Apple Press is, well, unimpressed. (9to5Mac)
  
  
  
• OpenAI is working on its own AI wearable device, probably earbuds. (Tech Crunch)
  
  Well, unlike an AI pin which is not useful at all, earbuds are at least useful for cleaning earwax.
  
  What? Something different? Don't be silly.
  
  
  
• Apps for boycotting American products surge to the top of the Danish app store. (Tech Crunch)
  
  Nobody tell them.
  
  
  
• Microsoft is building 15 datacenters in Mount Pleasant, Wisconsin. (Fox6Now)
  
  This is on a site originally slated for a Foxconn factory, a plan that never went anywhere. Microsoft purchased the land from Foxconn and the first two datacenters are under construction now.
  
  
  
• Rocket company Blue Origin's TeraWave satellite constellation plans to deliver speeds of up to 6 terabits per second, bidirectional, to enterprise customers. (The Verge) (archive site)
  
  Which used to be a lot.
  
  
  
• Why is America increasingly obsessed with prediction markets that perform no better than chance? (The Atlantic) (archive site)
  
  They should read serious journals like Fascist Quarterly The Atlantic. We reliably get everything wrong.
  
  
  
• AI company Eightfold is being sued in California for - apparently - successfully removing litigious idiots from lists of job applicants. (Reuters)
  
  Also, there's this self-ad in the middle of the article:

  Make sense of the latest ESG trends affecting companies and governments with the Reuters Sustainable Switch newsletter. Sign up here.

  No. I don't think I shall.
  
  
  
• Ubisoft has ended remote work and told staff to be in the office five days a week because it makes mass layoffs more enjoyable to watch easier to manage. (Notebook Check)
  
  If you had bough Ubisoft stock for 100 Euros in 2018, you would have jumped off a building by now.

posted by Pixy Misa at 04:00 AM

| Access Comments