Where we’re going, we don’t need chatbots
Read the original on dbushell.com ↗“You’ve been working hard, why not take a weekend vacation!”
Your chatbot says, after a 72 hour unattended session. It says that New York is nice this time of year. You ask it to book the cheapest flights. Half the airlines respond with a 429 code because your residential IP is part of five botnets. That’s odd you think, you only signed up to two that you can remember.
For the other airlines, your chatbot was blocked by CAPTCHAs at all but one, where it successfully identified a pelican crossing on the six-hundredth and seventh loop. It tries to buy tickets but the React web UI silently errors.
Unperturbed, your chatbot discovers a Github repo with API keys for Fedex. You’re booked as freight and must delivery yourself to the nearest depot 100 miles away. Based on your biometric data, your chatbot has estimated that you’ll comfortably fit into two 30×30cm boxes. You ask your chatbot to cancel the flight and rebook a business class ticket. It confirms that is has successfully cancelled the flight. A month later a Fedex invoice is emailed, but Outlook is down so the email bounces.
In New York, whilst attempting to order a pistachio matcha latte with an avocado and kale bagel (mashed, not diced, gluten free on the top half) from a food cart chatbot in Central Park, you get talking to another chatbot enjoyer. The conversation gets boring when they started rambling on about the funny things their chatbot said. Looking for an excuse to leave, you ask your chatbot pendant what appointments you have later today. It says it can’t establish a Bluetooth connection to your phone. You excuse yourself anyway saying: “I’ll have my chatbots contact your chatbots.”
Your chatbot open a Slack channel with the stranger’s chatbot and weeks are spent deliberating the intended purpose of this connection. Just as an hypothesis based on the 1998 rom-com You’ve Got Mail seems promising, you notice the errant session and kill it.
Your chatbot pendant has run out of battery but your watch chatbot, after finally processing the previous request, alerts you of the audio immersion tour it booked at Ground Zero. You ask it to book a driverless Uber but when it doesn’t respond you flag down a minicab.
You spend a poignant hour listening to the synthesised voice of Donald Trump read the Grokipedia page for 9/11. Somewhat skeptical of the source, you start a deep research chatbot to corroborate the facts. Your research chatbot cannot find any information for November 9th and writes its conclusion to dev/null. As you reach the memorial, chatbot Trump reaches the climax of his audio tour by blaming Biden and directs you to exit through the gift shop where they’re selling two-for-one golden Twin Tower merch. You ask the gift shop chatbot why one of the towers looks like Orthanc at Isengard. The chatbot explains a translation error when placing the order with its Chinese counterpart.
Back at Central Park, a homeless man wins an argument with an automated food cart over the culinary values of kale. He was absolutely right. An Uber car arrives and he steps inside.
Whilst walking around Times Square, an embedded billboard chatbot recognises your face and reminds you that you haven’t eaten today. You ask the chatbot to make a reservation at a restaurant recommended by Google’s AI overview. The first recommendation is fully booked years in advance due to a prompt caching error. The second recommendation turns out to be a fish market. The third recommendation is Dorsia, which you’re pretty sure is a fictional establishment from the 2000 horror film American Psycho.
You settle on the fourth recommendation describe as “Greek-Mexican fusion”. This delightful family run eatery advertises with gen-AI imagery of Tortilla wrapped Gyros and Halloumi Tacos in the style of Studio Ghibli. You order the Bean Moussaka from the waiting chatbot which compliments you on your exquisite taste. The chatbot reads the soft drink menu and you decline each option. You repeat the process for the hard drink menu, the cocktail menu, the coffee menu, and the soft drink menu again, before asking for a glass of tap water. The chatbot compliments you on your exquisite taste.
You finally arrive at your hotel and the reception kiosk chatbot is out of order. Luckily, its final act appears to have been to release all keycards from the automated dispenser. You fish through the pile spilled on the hotel lobby floor and find your keycard printed with your full name, address, and passport number.
You enter your hotel room and command: “Computer, lights!” After thirty seconds of no response you manually switch on the lights. The hotel room chatbot then replies: “Aye, captain.” and turns off the lights. The lights in your apartment back home also flicker.
You climb into bed and unfold your new travel USB multi-charger onto the bedside table. You place your pendant chatbot into the lower socket, hang your watch chatbot on one arm, and rest your phone on the central platform. The multi-charger chatbot asks if you would like it to slow charge at off-peak hours. You ask it if the hotel has any hidden fees for electricity usage. It requests root permissions and begins to recite terms and conditions. You ask it to stop but it’s no longer listening so you move it into the bathroom where it continues to read terms in multiple languages all night.
You doomscroll X for an hour until you read a horrifying story of an Uber Tesla catching fire and burning a man to death inside. According to witnesses, the first responder chatbots refused to help citing AI safety guardrails and the dangers of fire, advising to dial 911 in an emergency. Several bystanders attempted to redial, only to hear the same message.
You ask your phone chatbot to set an alarm for 7am. You hear the faint voice of Siri from the adjacent room saying that is has ‘found some web results for “alarm for 7am.”’
Thanks for reading! Follow me on Mastodon and Bluesky. Subscribe to my Blog and Notes or Combined feeds.