Monday, March 27, 2023

Tech news bits: ChatGPT improves rapidly; Cat videos are crippling war material supplies

This post ushers in a whole new universe of content here at DP, tech stuff!! Yeah, well, tech stiff is not new here. Whatever.

From the Computers Are Going To Take Your Job Files -- ChatGPT rapidly gets a lot smarter: One could have reasonably believed that ChatGPT would get better over time. But this fast. Insider writes:
A professor says he's stunned that ChatGPT went from a D grade 
on his economics test to an A in just 3 months

An economics professor said the progress ChatGPT made — it improved its score from a D to an A on his economics test in just three months — has stunned him.

when ChatGPT-4 debuted, its progress stunned Caplan. It scored 73% on the same midterm test, equivalent to an A and among the best scores in his class.

ChatGPT's paywalled upgrade sought to fix some of the early issues with the beta version, GPT-3.5. This purportedly included making ChatGPT 40% more likely to return accurate responses, as well as making it able to handle more nuanced instructions.

For Caplan, the improvements were obvious. The bot gave clear answers to his questions, understanding principles it previously struggled with. It also scored perfect marks explaining and evaluating concepts that economists like Paul Krugman have championed.
__________________________________________________________________________
__________________________________________________________________________


From the Sucking The Juice Out Of The War Files: Insider writes:
Weapons firm says it can't meet soaring demand for artillery shells 
because a TikTok data center is eating all the electricity
  • An arms manufacturer complained that TikTok's data center is using all the electricity in the region.
  • As a result, the company cannot keep up with the surging demand for artillery rounds.
  • The CEO told the FT the company's growth is "challenged by the storage of cat videos."
"We are concerned because we see our future growth is challenged by the storage of cat videos," said Morten Brandtzæg, the CEO of the Norwegian arms manufacturer Nammo, in an interview with the Financial Times.

Brandtzæg said the demand for artillery rounds was 15 times higher than normal — a trend driven by the war in Ukraine, which has featured heavy artillery use.

Ukraine, for instance, would like to increase its daily usage of rounds from 6,000 to 65,000, he said.

Ukrainian soldiers fire at Russian positions from a U.S.-supplied M777 howitzer 
(my favorite artillery piece) in Kherson region, Ukraine, Jan. 9, 2023

Actually, multiple sources are reporting that there is a huge gap in what US arms manufacturers can produce and the vast demand for arms and supplies from Ukraine. Lots of stories had come out saying that the Russian could not keep up with demand, and so far that has not made much noticeable difference. But the US military is now reporting that its reserves of all kinds of modern arms and ammo are being rapidly depleted and manufacturers cannot keep up.

Does anyone else get a sense that the wheels have come off the cart in America, or is that just my paranoia speaking? Remember when America could not even make simple cloth. M95 and N95 masks in the early days of COVID? America feels significantly broken and mostly not fixable, especially if brass knuckles capitalism gets much more control of commerce and government than it already has.

__________________________________________________________________________
__________________________________________________________________________


Making an updgrade by downgrading it: Insider writes on a trend that many consumers are noticing. If I recall right, I first started noticing this about 25 years ago.

Google, Amazon, and Meta are making their core products worse — on purpose
In recent years, Google users have developed one very specific complaint about the ubiquitous search engine: They can't find any answers. A simple search for "best pc for gaming" leads to a page dominated by sponsored links rather than helpful advice on which computer to buy. Meanwhile, the actual results are chock-full of low-quality, search-engine-optimized affiliate content designed to generate money for the publisher rather than provide high-quality answers. As a result, users have resorted to work-arounds and hacks to try and find useful information among the ads and low-quality chum. In short, Google's flagship service now sucks.  
All of these miserable online experiences are symptoms of an insidious underlying disease: In Silicon Valley, the user's experience has become subordinate to the company's stock price. Google, Amazon, Meta, and other tech companies have monetized confusion, constantly testing how much they can interfere with and manipulate users.
This makes perfect sense if one looks at it from the point of view of a brass knuckles capitalist. Why try to improve something that already works, when it is easier to break it and add some useless fluff to distract from what was broken? Why not stuff ads and sponsored links in a search result if it forces consumers to wade through the garbage to find what they want, thereby generating some extra profit when an ad snags someone?

I notice this most acutely now with Amazon, where searches are now loaded with crap I don't want and did not search for. Unfortunately, Amazon does not have much in the way of significant competition. When I try a non-Amazon source, the selection is usually poor and/or service bad or non-existent. 

__________________________________________________________________________
__________________________________________________________________________


AI and human intelligence are becoming a gray area?: Vice writes about an alleged blurring of the line between humans and machines:
On Wednesday, Microsoft researchers released a paper on the arXiv preprint server titled “Sparks of Artificial General Intelligence: Early experiments with GPT-4.” They declared that GPT-4 showed early signs of AGI, meaning that it has capabilities that are at or above human level. 

“We demonstrate that, beyond its mastery of language, GPT-4 can solve novel and difficult tasks that span mathematics, coding, vision, medicine, law, psychology and more, without needing any special prompting,” the researchers write in the paper’s abstract. “Moreover, in all of these tasks, GPT-4’s performance is strikingly close to human-level performance, and often vastly surpasses prior models such as ChatGPT. Given the breadth and depth of GPT-4’s capabilities, we believe that it could reasonably be viewed as an early (yet still incomplete) version of an artificial general intelligence (AGI) system.”  
Microsoft researchers observed fundamental leaps in GPT-4’s abilities to reason, plan, solve problems, and synthesize complex ideas that signal a paradigm shift in the field of computer science,” a Microsoft spokesperson said.  
With all this being said, it is clear that the “sparks” the researchers claim to have found are largely overpowered by the number of limitations and biases that the model has displayed since its release.
Although the paper's title is asserted to be more click bait than substance, the technology is improving rapidly. At some point fairly soon, there is going to be serious blurring of work by AI and humans in increasingly complex topics. One thing that makes improvements faster and better is learning what limitations and biases the AI has. Once those are known they can be improved and fixed.

One question this raises is what about jobs and job losses. A poll conducted 10 days after the first version of ChatGPT came out indicated that over half of employers who tested it, had started to lay people off and replace them with the AI software. Other questions are how far will AI make inroads into jobs and and how fast will it happen? 




No comments:

Post a Comment