Category: Dragon Musings

Random thoughts, experiences, and reviews on anything not related to food, drink, or travel.

Travel Tips from the Edge

So I just came back from a three-week vacation to Scotland, Northern Ireland, and the Republic of Ireland. Now I need a vacation from our vacation.

Over the coming weeks I’ll post pictures and musings (Scotland and Ireland are both incredibly beautiful countries), but for now I just wanted to share a few tidbits I learned (and perhaps I should’ve known) about vacation planning and execution to international destinations.

Most importantly, do NOT book airline travel through a third party such as Travelocity. They do not actually do anything except book your flights and blow hot air. More on this in a minute. As a corollary, with perhaps a couple of notable exceptions, it’s best to book with an airline domestic to your country of origin. “Codeshare” is merely a marketing ploy and doesn’t actually mean the booking airline has any knowledge or control over your connecting flights in a country where it’s not a domestic carrier. It’s merely a convenient way to market an international destination.

The Specifics

The trip: Miami- Dublin, Dublin-Miami. Airlines involved: American Airlines for Miami-Philadelphia and United Airlines for Chicago-Miami. Aer Lingus for Philadelphia-Dublin and Dublin-Chicago. (No one flies direct to Ireland from Miami.)

My Tale

After much research, I booked our flights to and from Ireland through Travelocity because ostensibly we’d save @ $400. However, by the time I was done paying for seat selection and luggage, I actually paid @ $100 more than if I’d just booked directly with Aer Lingus. Why? Because of the myriad “economy” designations airlines use today. All economy designations are not created equal. My best guess is when you purchase through these online only travel agencies (a misnomer-in my opinion, they’re actually just resellers), the airlines factor in their reduced profits and adjust accordingly.

Monetary considerations aside, my saga continued. In my scenario, you must still contact the individual airlines directly to select/purchase seats and/or prepay luggage, which I did. Since I booked our travel @ six months in advance this meant every time there was a change in the flights, I had to contact each airline to transfer our seat selection as this does not happen automatically. There were three flight changes and three airlines involved before we left. Wait times varied but I calculate I spent at least 6+ hours on the phone making these changes since it’s not something that can be done electronically.

*As an aside, at least with Aer Lingus, if you book through a third party you lose the ability to check in or access your boarding pass through their mobile app, which means check-in at the airport requires a stop at their customer service counter (with its attendant long lines).

A week before our scheduled departure I logged in to confirm our flights. I started with Travelocity and confirmed there were no further changes to our itinerary. However, since they don’t track your seats or prepaid luggage selections, I logged into the individual airlines to confirm the particulars.

The plot Twist

This is when United Airlines informed me the flight we were originally booked on from Chicago to Miami was cancelled. They said we had a credit and they were waiting on Aer Lingus to tell them whether to refund or put us on another flight. Since it was a “code share” with Aer Lingus further complicated by the third party reseller, I did not have the authority to make the change myself.

I contacted Aer Lingus, whose representatives told me I was misinformed and the flight was confirmed. I contacted Travelocity who said much the same but promised they’d look into it and get back to me within 72 hours. I never heard back from them (to this day) and we left with this issue unresolved.

I decided to cross the bridge when we arrived back in the US and enjoy our vacation. When we checked in with Aer Lingus in Dublin for our return flight to Miami via Chicago, they still confirmed the non-existant flight to Miami and told us our luggage would be checked through to our final destination and our boarding passes for the next flight issued in Chicago. Interestingly enough, the luggage claim checks we were given listed a different flight number for the Miami flight than the one they claimed we were booked on (but I didn’t notice this until we landed in Chicago).

We land in Chicago and we know we need boarding passes. The Aer Lingus gate agent tells us to proceed to the United terminal and they would issue our boarding passes at the gate. I press him on the details but he dismisses me with an indifferent “we don’t fly domestic” and tells me United will take care of it. We have four hours until the supposed flight so I know without needing to check there will be no gate assigned yet (at least not one posted for the public) and that there certainly will be no gate agent responsible for our flight at the as yet undetermined gate (for the imaginary flight). But there’s really no point in arguing with him so we start to make our way to the appropriate terminal for United.

I thought of calling Travelocity itinerary support but we didn’t want to sit in Chicago for three days waiting for them to get back to me-which is when I examine the luggage claim checks more closely and discover the discrepancy in the flight number. While we’re walking, I check the United app and note the flight number on our luggage is actually from a flight to Chicago from Detroit that already landed. By now, I wouldn’t be surprised if our luggage is actually on the way to Dubai but I check the airtags and confirm they arrived with us in Chicago.

We stop at the first United counter we come across. It’s cordoned off and in order to be allowed to speak to a customer service rep you first have to explain to the cordon guardian why you can’t simply use the automated kiosk. I kid you not. I was only about 1/3 of the way through my explanation when he finally let us through, after confirming the discrepancy between our luggage claim checks and our final destination.

The United agent at the desk confirms what I knew a month before, we’re not actually booked on any flight to Miami. But she finds our previous reservation and corresponding credit and tells us there’s room on the 6:30pm flight to Miami (which was more less the same time of the original cancelled flight). Through yet another mistake of Aer Lingus or Travelocity (who knows), we were booked first class originally so they put us in first class back to Miami. (I figure at this point we’re owed so mum was the word.) But, she says, there’s nothing she can do about the luggage because Aer Lingus hasn’t transfered it to United yet. We could only hope once the luggage was transfered to United and they discovered the discrepancy they’d put it on the right flight. Otherwise we’d have to file a claim at the Miami airport for our wayward luggage.

The Happy (relatively) Ending

By the time we arrive in Miami at 1:30am we’ve been awake and traveling for almost 24 hours and I’m resigned to maybe seeing our luggage sometime the following week (and uncaring as long as I get to sleep in my bed), but low and behold, when I check the airtags, our bags are indeed sitting in Miami with us. There were no new tags on them, but somehow they were put on the right (only) flight to Miami. So all’s well that ends well I guess.

Lesson Learned

I will never again use third-party resellers or online only travel agencies for international travel. They’re a good source for exploring itineraries and getting an idea for flight prices, but they’re lousy at actually providing any customer support when you need it. We could’ve flown United, for example, for all legs of the trip for the same price and saved time and most importantly preserved our peace of mind during our trip.

AI🔗Trust & Safety

Behind the Curtain

I consider myself a discerning user of technology. I have an Echo (aka Alexa) but not a smart refrigerator. The first makes my life easier, the second would tip me over the edge. So although I’m not in the field, I recognize that we’re living in “interesting times” technologically speaking.

In my lifetime we have seen the advent of the PC, ATMs, MRIs, cell phones, the internet, DNA testing, barcodes, GPS, bluetooth, hybrid cars, iPhones and the one I most love to hate, social media-to name just a few of the most prominent innovations that have absolutely and irrevocably changed the way we live our lives.

I’m also an avid reader, so when AI was thrust so prominently into the spotlight, particularly so dramatically in the last year, it was no surprise to me. The dumb money (to borrow a financial term from another life) was finally catching on.

AI was first developed in the ’50s and is already being used more extensively than the average person realizes or thinks about (at least until they need customer support for, well, anything really). The only thing holding AI back before was computational power. But it was turbocharged in 2020 with several breakthroughs in neural network design. Welcome to the age of big data.

So yes, I agree with those who call AI our age’s industrial revolution. There’s no going back and it will change everything and there will be growing pains and a necessary realignment of the workforce. It will bring positive advances and negative, sometimes unintended and unforseen, consequences.

So in thinking about AI two interralted issues concern me the most, as an average American. What data, exactly, is being used to train these AIs and who’s monitoring their output to at least sound the alarm when inevitably something goes wrong?

Right now, we’re at the frontier of large language models, which are the types of AI that mimic human intelligence. I’ll explain more about how these AI are trained in the second part of this post, but basically vast amounts of data are fed into the models so they can learn the patterns and connections between words and phrases. The theory being the more data used, the better the software will be at generating new content.

But common sense, and a bit of life experience, tells me garbage in, garbage out. Computer scientiests and mathmeticians even have an acronym for it, GIGO. But I don’t see much discussion of it in mainstream media.

Who’s making sure the companies developing these LLMs are actually using high quality data as opposed to feeding the models every bit of garbage that comprises the modern day internet? I know who’s not doing it…the governement. Which leads to my second interrelated concern.

Specifics vary across fields and companies, but basically trust and safety departments exist on online platforms to protect their users, whether from fraud, harassment, offensive content, spam, or misinformation and disinformation.

As someone concerned about the state of American democracy, that last one concerns me the most, specifically on social media sites. A democracy cannot survive in the absence of a shared reality based in a factual world. The only thing today maybe we can all agree on is our society is fractured. IMHO misinformation and even more so disinformation will only widen those fissures.

Recently the head of Twitter’s trust and safety team resigned and pretty much every major news outlet has reported on how tech layoffs have affected trust and safety departments leading to fears of backsliding on efforts to curb online abuse. (Google it.) So who’s watching out for that stealth bolt quietly spreading the next conspiracy theory on social media?

QAnon started on 4chan and in no time exploded to mainsteam social media sites and now it’s as popular as some major religions. And they didn’t have the easy access to chatbots available right now. (More on chatbots below.) The left is not exempt from its own share of conspiracy enthusiasts either with sites like the Palmer Report, Shareblue, and Patribotics. The danger has disinformation researchers in a tizzy and led to the man commonly called the “Godfather of AI” to resign from Google.

So right when chatbots are poised to become not only much smarter, but also better at mimicing human speech patterns, social media sites are downsizing the very people poised to keep them in check or at least let us know when something’s amiss? Even in the era of “move fast and break things” this seems like a sure fire recipe for disaster.

Cliches and truisms are oft-repeated because sometimes they just say it best and right now one comes to mind. Bells cannot be unrung. At this point no one is going to dissuade QAnon believers, election deniers, or even moon landing skeptics their beliefs are factually incorrect. I’d rather not see what kind of conspiracy theory an intelligent AI could quickly proliferate at the behest of a bad actor, foreign or domestic.

The Deep Dive

Before I get into the next part of my post I want to explicitly state something that’s probably obvious if you’ve made it this far: I’m a bit of a geek. But if you’re still reading, you’re probably a bit of a geek too. I have read many scholarly articles on AI and chatbots, and large language models from sources I consider credible, mostly peer reviewed scientific journals and niche publications. Whenever I’m interested in a topic I tend to do a deep dive and learn more than any of my friends and family are willing to listen to. You can find some, though not all, of them here.

chatbots

Chatbots have been around for a long time but really came into popular use in the late ’90s. The first one, Eliza, was created in 1966 by an MIT computer scientist. ChatGPT and BARD are the two most talked about right now, but generic chatbots, essentially computer programs that can interact and respond to humans in real-time “conversations” are literally everywhere in customer service departments gathering data and routing calls.

Early on, it was easy to tell that you were “speaking” to a computer. Something that I think is lost in a lot of the noise over the very real capabilities of modern AIs is that what we’re really talking about is computers that can imitate humans convincingly. I’m simplyfying, but if you’re interested in the underlying philosophy and scientific priciples, see here.

Large Language Models (LLMs)

Neural networks are a type of machine learning model that are loosly based on the human brain and the heart of deep learning algorithims. These algorithms are how AIs are trained. Large language models are basically computers that understand and generate text in a human like fashion.

They are trained using a technique called “deep learning”. This involves feeding massive amounts of text data into a neural network with multiple layers, allowing the network to learn the patterns and structures of language. Language models are trained on vast amounts of text data because, unlike humans, they don’t have innate knowledge of language and must learn from scratch. The training data provides the model with a vast range of linguistic patterns to learn.

Jetpack AI

The quote above was generated by the Jetpack AI availble right now on WordPress to subscribers. I asked it how large language models are trained and that was the initial part of its response, which I edited so as not to put anyone to sleep.

When I asked it where the data comes from, it responded as follows:

The text data used in deep learning is often obtained from a variety of sources such as publicly available datasets, web crawls, or data partnerships with companies. While some large language model training datasets are curated by academic or industry organizations, in many cases the data is sourced or purchased from a variety of sources. In general, no one entity or organization controls all the text data used in deep learning.

Jetpack AI

When I asked it if hate speech was used in the data fed into the large language models it told me this:

The use of hate speech in deep learning is a controversial topic, andit is up to individual data providers and researchers to decide whether or not to include it in their training data. Some argue that limiting the training data to exclude hate speech will result in more ethical and responsible language models, while others argue that omitting such data will lead to models that are less generalizable to real-world scenarios.

However, it is important to note that hate speech is not condoned and has no place in society. Large language models can help in identifying and mitigating hate speech by using natural language processing techniques to better understand the context in which hate speech occurs and develop more accurate detection algorithms.

Jetpack AI

Of course if no one is spending the money to employ AIs to monitor hate speech or disinformation the point is moot. But as you can see, the capabilities of the technology are quite remarkable and easily accessible to anyone already.

Now, I’m not a coder or a hacker or a programmer, but its extremely possible and plausible that anyone with those skillsets could program and deploy a chatbot to spread hate or disinformation on social platforms. It’s already happened. It’s why Microsoft had to shut down its Tay chatbot after only a few hours on Twitter in 2016, when it started spewing offensive and hate-filled tweets “learned” from a post on 4chan.

Hence why I’m concerned about the use and proliferation of these chatbots and the lack of oversight. In my opinon it’s already impossible to tell whether you’re actually chatting with a person or a computer online.

Just in the last year, a fake twitter account caused stocks to plummet for Eli Lilly when it announced free insulin, a lawyer used ChatGPT to create a brief only to find out later it made everything up after he’d filed it in federal court and BING, Microsoft’s AI chatbot, threatened users.

I think its important we’re informed so we can govern ourselves accordingly.

Legos

Lego Technic Jeep Wrangler

I loved playing with legos as a kid and now my oldest niece loves them too. But her legos are surely not the ones of my youth. My sets were simply humdreds of lego pieces in different colors, shapes, and sizes and the instruction book was my imagination. (I was delighted when they added doors that opened!) Her sets have wheels and moving pieces and come with instruction books with designs devised by NASA engineers.

I wouldn’t go so far as to say the sets stifle creativity, but I’m not sure she gets the same level of fun out of them as I remember. She’s a gifted kid, well on her way to becoming an overachiever, and when she misses a step she gets frustrated and discouraged. Some of the sets, even for younger kids, require hours of build time and can take days to complete. I’ve had to bail her out more than once. Hedwig and Yoda were memorable, to name just a couple.

When I played with them as a kid I just built whatever inspired me in the moment and played until I was dragged away. I always looked forward to when I could play with them again and feelings of frustration, disappointment, and failure were never ever associated with my beloved legos. Maybe I’m just getting old. At some point I think every generation thinks “things were so much simpler” when…(fill in the blank).

My experiences with her, however, did inspire me to build my own. I recently sold my sunny colored stick-shift Jeep Wrangler. I loved that car and I drove it for over 15 years, but driving stick in a crowded city sapped the life out of me and it had to go. But I miss it. I miss walking out of my house and seeing it sitting in the driveway, bright yellow and happy, no matter what kind of day I was having or what the weather was like…*sigh*.

But at least now I have a bright yellow reminder of all the joy it brought me for over a deacade and a half. It sits on my desk with its light kit that I turn on from time to time while I reminisce.

So my sincere thanks to the Lego Group even if I feel they might have lost a bit of the leg godt (Danish for “play well”) philosophy that insprired it all.

Cloud 5 Shoes

I recently bought a pair of OC Cloud 5 Waterproof lifestyle shoes in Cocoa/Frost. OC is a company based in Zurich that employs Swiss engineering to design their shoes. They have a company philosophy of sustainability and limiting their corporate footprint on the planet.

Cute, right? The shoes too. (That’s my dog, Stella, unimpressed as ever.)

They came to my attention when I was doing research for a walking shoe for an upcoming trip. They made the cut on several lists for best walking shoes. I am a true US size 8M in shoes. Whenever a size 8M shoe doesn’t fit, there’s usually no fix because the next size up will be too big.

The day I received my shoes I tried them on. They were tight all around, length and width. I sent them back the same day, but decided to try the 8 1/2 because they were really cute and recommended by so many different respected groups. (I don’t mean review sites, I mean “best of” lists by walking and hiking publications and groups.) Thus began my ordeal.

Somehow the shipping label got mixed up when they sent me my exchange pair in the new size, and instead of delivering the shoes to me, they went to OC’s office or warehouse, or who knows what location. Fine, mistakes get made. But it wasn’t easy to straighten out the issue or get ahold of customer service and have someone actually figure out what happened and where my shoes were because they’re so automated, I kept getting emails saying my shoes were on the way or delivered already.

Over 4 weeks passed and I didn’t have my shoes or a refund. They’re not cheap shoes. I was angry and frustrated. I disputed the charge with my credit card company and coincidentally (not!) the very next day someone reached out from the company and explained what happened.

They offered me a discount for the hassle and I agreed. I finally had my size 8.5M a couple of days later. Then the real pain started, literally.

I’m going on my trip in August and I was researching shoes early so I could get the “breaking in” over with before then. I wanted them for urban walking. The first day I tried them out on a walk around my neighborhood. They felt a little snug, but I thought they might give a little after some use. They caused horrible blisters on my Achilles tendon.

I persisted because at this point they couldn’t be returned since they were used. I wore them daily on walks for over three weeks before they finally ceded enough to be comfortable without causing any blisters, but they’re still not great.

Now, in all fairness, the company calls this model a “lifestyle” shoe, not specifically a walking shoe, but unless your lifestyle consists of only walking from your destination to your car and back, these shoes won’t cut it. I’ll use them if I’m hanging out at a bar or a friend’s house or something, but they’re not suitable for a day of running errands or hitting the mall. I honestly think there’s some design flaw or dimension mistake in their manufacturing. Maybe it’s just this model or something happened with this color because I noticed they’re no longer offered in the cocoa/frost on their website. Who knows?

Lesson learned for me. No more mail order shoes or accepting any size but my tried and true. Several people have recommended the Hoka line of walking shoes, including my primary physician, so we’ll see how that goes. I’ll be off to the sporting goods store soon.