ChatGPT and Congress

 Yesterday there was a report, which I may be garbling, that Google had given ChatGPT the same test questions they give to engineering job applicants, and the AI qualified as a level 3, apparently an entry level.  The starting salary for level 3 was given as about $180K, more than the starting salary for a new member of Congress, not to mention a member of considerable seniority. 

Not sure what that says about AI, Google, Congress, or the US. 

The Answer Is Google, Always Google

Supposedly intelligent people still aren't current with the modern world.  Two instances:

1. Mr. Kushner tried to find out the name of the Russian ambassador to the US (that's in the Mueller report) in late 2016.  So he called someone who might know.  
2. Scott Adams tweeted out a reward of $100 to the first person who could tell him how to change the footnotes in a Word document from "i, ii, iii..." to "1, 2, 3".

In both cases simply typing the question into Google would have produced the answer in a matter of seconds.

I hope our young are learning this lesson better than their seniors (both of whom could be my children, God forbid).

AI, Dark Swans, Google Map Directions, and Moore's Law

Some further thoughts on the Google map episode I described yesterday.

It strikes me that AI in general will have problems with "dark swan" events.  That's true by definition--if AI is trained by using a big database of past information, it can't be trained to handle events which aren't reflected in the database.  In many cases, like autonomous cars, the algorithms can be set to do nothing.  It would be similar to the "dead man switch" found on locomotives--if there's no longer an intelligence at the controls which can react properly to events, you stop the engine.  But in other cases it may be harder to define such cases.

In cases like I encountered yesterday there may be salvation found in Moore's Law.  Presumably Google samples current traffic volumes using some priority rules--sample most often the most traveled routes.  But priorities are needed only when resources are scarce; as technology becomes cheaper it would be possible to sample everything all the time (which is almost what the human sensory system does).

Google's Driverless Car Comes to Washington

Here's a Politico report on the car's success in navigating Capitol Hill streets, not the halls yet. It includes this question:

But there are questions about its viability — will consumers buy into losing control behind the wheel?

Read more: http://www.politico.com/news/stories/0512/76459.html#ixzz1vEYR7joD

Speaking only for myself, I could see buying it--I'm nearing the point where my self-confidence is my driving ability is starting to fade, so an old geezer I've love the ability to delegate 99 percent of the driving to a computer.  I suspect that's how the innovation will come; something like Segway which was promoted as revolutionary but has turned out to be a niche filler.  Between geezers and drunks there's a big niche to fill.

Easy Data Collection

Megan McArdle alerted me first to the possibility, then there was a more elaborate post here.

What's the possibility?  Using a Google Docs spreadsheet/form combo to capture data.  In McArdle's case, she's using it to capture the name and addresses for her wedding invitees.  As a bureaucrat, I can't help but remember a number of occasions when I would have liked an ad-hoc report from state offices.  But trying to get Kansas City to gin up a short report was difficult.  Granted, in many cases the data I wanted was sitting in the county files and could be extracted by using Query/36 and uploaded, but there were some cases where asking the office to data load something would have been helpful.

Now, thanks to Google, it's available.  Or, to be realistic, it would be except undoubtedly the FSA hierarchy will put some limits and restrictions on using Google, like: "Do not ever use it."