Another Month of What People Are Doing with Artificial Intelligence

The users are finding more and more applications for the most newsworthy technology of the 2020s.…

Nikolas Lanum, in Fox News on March 22^nd, told us that a “Texas private school’s use of new ‘AI tutor’ rockets student test scores to top 2% in the country.”  At Alpha School, “students are placed in the classroom for two hours a day with an AI assistant, using the rest of the day to fucus on skills like public speaking, financial literacy, and teamwork.”  Their ability to work on things they call “passion projects,” and therefore have strong interest in, can explain these results, but for that AI may indeed be the way to go.

We saw also that “AI enables paralyzed man to control robotic arm with brain signals” (Kent Knutson, Fox News, March 30^th).  The experimental achievement started with “sensors implanted on the surface of his brain,” which “recorded neural signals as he imagined movements like grasping or lifting objects,” and “over two weeks, these signals were used to train the AI model to account for daily shifts in brain activity patterns.”  After months of practicing “controlling a virtual robotic arm,” followed by using a “real” one, “he quickly mastered tasks such as picking up blocks, opening cabinets and even holding a cup under a water dispenser.”  Promising, but will take years at best before it can be provided at scale.

The same source told us about “The dangers of oversharing with AI tools” (April 9^th).  While the likes “of ChatGPT have become incredibly adept at learning your preferences, habits and even some of your deepest secrets,” that means their “knowing” so much about you “raises some serious privacy concerns.”  Information they have may be relayed back to their manufacturers, but it is not clear how much damage that actually does.  It may take a known case of someone being badly hurt before we can effectively regulate, or even understand, the true threat.

More tamely, “These AI transcription voice recorders surge in popularity” (Christopher Murray, still Fox News, April 19^th).  With them, users can “record, transcribe and summarize content effortlessly,” using AI’s ability to “transcribe in 112 languages” and “generate comprehensive summaries” – now.  One, PLAUD’s NotePin, is wearable as a pin, a wristband, or a necklace, and, by using encrypted cloud storage, presents no privacy concerns.  The devices are generally low-priced but require paid subscriptions for the AI itself.  This is one use of AI almost certain to continue without existential scariness.

Could we say the same about coworkers who aren’t real?  On that, “Anthropic anticipates AI virtual employees coming in next year, security leader says” (Alex Nitzberg, Fox Business, April 22^nd).  That company is creating “digital AI employees” with “”memories,” parts to play within the business, and company accounts and passwords,” with “much greater autonomy than agents currently do now.”  Yet even Anthropic’s chief information security officer of the title said there are many such unsolved problems, such as “AI employees” being able to “hack the system in which code is merged and tested prior to being rolled out.”  Since this is more of a potential application than a real one, it may not belong in this post at all, but since users can clearly now create bogus employees of some sort, with small but growing capabilities, we should be aware that new names on organization charts may not be Homo sapiens – or anything alive at all, even if they can converse.  Yes, it may not be possible, as author Erle Stanley Gardner once had his character Perry Mason say, to “correspond with a corpse,” but it is with these electronic automata.  So don’t be fooled – or surprised – and that goes for the other four AI functions here as well.