AI and crap detection

For nearly two decades, I've been writing and teaching about the skills and mindset necessary to find the true online signal amid the abundant noise. As soon as my daughter started using infoseek and other pre-Google search engines when she was in middle school, I started showing her how to verify the results of web searches much later, (she later worked at Google for 8 years). I started publishing videos about crap detection in 2009 and it was one of the essential social media literacies I wrote about in Net Smart (2012). Two trends have become prominent in recent years: Education has failed, over the decade since Net Smart, to prepare students anywhere in the world for living in a world of abundant badinfo; at the same time that AI opens new capabilities for augmenting thought and sense-making, it also supercharges the machinery of fakery. It used to be akin to finding a needle in a haystack; increasingly, 

I am increasingly convinced that crap detection skill is the great divider of coming years: You can use search to answer any question,any time, anywhere, but you have to find the signal amid a lot of noise -- much of which is now deliberately generated to be hard to distinguish from signal.  (More and more, the task is akin to finding a needle in a haystack full of fake needles) A good new book, Verified, inspired by my earlier work, is an excellent contemporary guide to web verification. 

Now with AI chatbots, instant knowledge, graphic and video-making capabilities are available -- if you know how to dig the signal out of an ever-growing tsunami of bullshit, disinfotainment, hallucination, computational microtargeted propaganda. Considering that more than half of adult Americans read at sixth grade level or lower and 20% can't read at all,  and considering how educational institutions have failed elementary crap detection education, I don't see the public sphere improving from where it is now. Individuals who master these info-skills will be empowered, but we are a tiny fraction of a population where the number of people who believe that earth is flat is increasing rapidly (Google "flat earth group).

I don't have enough basic understanding of how machine learning and large language models work to have an opinion on whether general artificial intelligence or an ai singularity are imminent. I'm inclined to think of AI chatbots as a "cultural technology," as Alison Gopnik addresses them, like speech, writing, mathematics, print, graphic interfaces to personal computers, the Web. Like those previous cultural technologies, interactive LLMs don't explain how to use them -- literacies are emerging from populations of users pushing the limits. I have been particularly interested in Gopnik's take because of her pioneering work on learning in infants. 

Here are two recent links, well worth reading, that inspired me to write this post:

AI has a terrible energy problem. It’s about to hit crisis point

That world would see us need perhaps ten times as many cloud computing facilities as we already have. Since cloud computing already consumes more than 2% of all electricity generated worldwide, we’d be looking at an AI future – within the next decade – where at least a fifth of all of the electricity we generate willbe going to power cloud computing installations."

A “shocking” amount of the internet is machine-translated garbage, particularly in languages spoken in Africa and the Global South, a new study has found. 

Researchers at the Amazon Web Services AI lab found that over half of the sentences on the web have been translated into two or morelanguages, often with increasingly worse quality due to poor machine translation (MT), which they said raised “serious concerns” about the training of large language models."