Artificial Intelligence Chatbots Are Here to Stay

ChatGPT has ushered in a new era for AI, though plenty of drawbacks remain.

By
16
In this photo illustration, the welcome screen for the OpenAI "ChatGPT" app is displayed on a laptop screen. (Photo by Leon Neal/Getty Images)

Buzzfeed is using them to make quizzes. Microsoft has invested billions of dollars in them even as it laid off 10,000 employees. Amazon is selling them to replace call centers. Artificial intelligence chatbots have been part of the business landscape for a while, but consumers should expect to see even more of them.

“I think it provides a promising approach for a kind of natural language interface to whatever you want,” University of California, Berkeley computer science professor Steven T. Piantadosi told The Dispatch. “If you want to be able to, I don’t know, book flights or order food or whatever just by talking to something, then it’s competent enough with language that it could form the kind of key interface between language and some other kind of system.”

By the end of 2022, 50 percent of respondents to a McKinsey and Company survey said they were using AI in their business in some form.

The big new player on the scene is ChatGPT, which parent company OpenAI launched as a prototype on November 30. (The “GPT” stands for “generative pre-trained transformer.”) The chatbot now has more than 30 million users, according to the New York Times, and it has made headlines for everything from passing a Wharton Business School MBA exam to emulating the deceased for conversations.. 

Keep reading with a free account
Create a free Dispatch account to keep reading Get Started ALREADY HAVE AN ACCOUNT? SIGN IN
16
By
Comments (16)
Join The Dispatch to participate in the comments.
 
  1. Avatar photo
    MatthewRobin

    "
    OpenAI CEO Sam Altman acknowledged in a tweet that ChatGPT “has shortcomings around bias,” promising that the company was working on it. ChatGPT gives users the opportunity to give a thumbs up or thumbs down to responses to ChatGPT, and Altman has asked users to downvote examples of bias the chatbot churns out to help it learn what is and isn’t appropriate.
    "

    This is worrisome because people's upvotes and downvotes influence chatgpt which encourages people to organize online to upvote ans downvote things in a coordinated manner.

    Also, no one is up voting or down voting based on the bias of chatgpt. The votes are reflective of the biases of the voter.

    Collapse
  2. Avatar photo
    The Bell Tower

    I think The Dispatch should do an experiment where they use AI to publish a news story using the byline of one of their writers and see how many of us pick up on it.

    Collapse
    1. Avatar photo
      Show Jay Janney the 🚪!

      Oh...So it's just a coincidence Nick came in about the same time as ChatGPT? 😮

      Collapse
  3. Avatar photo
    George.Skinner

    I'm not sure people understand just how different real AI is from the type seen in movies and TV. Imagine you've got a series of data describing some phenomenon like harmonic motion of a spring. You can find a trend two different ways. One is to describe the underlying phenomenon with a mathematical model that shows how the harmonic motion is produced. Another way is to fit a complex polynomial to the data. Either way is going to allow you to interpolate a point anywhere in your data set, and the fitted polynomial may even be somewhat better than the mathematical model at capturing the effect of various noises that cause the experimental data to vary from the model. However, if you change something fundamental like the stiffness of the spring, only the mathematical model can readily be changed to predict the new behavior. The accuracy of the fitted polynomial is tied to the data set it was generated for.

    Neural net AI is like that fitted polynomial, just vastly more complex. It's still orders of magnitude less complex than the neural nets of a human brain. That's the difference between a real-life Tesla that struggles to manage urban traffic and Nick Fury's self-driving car careening through a car chase in a Marvel movie; or a self-serve voice prompt menu for your cell phone provider struggling to guide you to the right option vs. Tony Stark's JARVIS making barbed quips as it solves complex problems.

    Collapse
    1. Joe1776

      You lost me at your second use of the word "polynomial."

      But I think I appreciate your point.

      One way you can understand sound is by listening to it. Another way is by measuring vibrations.

      Some people said Deeper blue was so smart for being able to beat Kasparov at Chess. Someone else pointed out that deeper blue didn't even know it was playing chess.

      Maybe your comments are addressing how that gap to consciousness is bridged. But it seems like it would be very difficult to get there and know if we got there.

      Collapse
      1. Avatar photo
        Show Jay Janney the 🚪!

        A polynomial is simply an unfaithful mononomial.

        Collapse
  4. Avatar photo
    Aylene Wright

    Thanks Alec for the balanced reporting that touches on the "ChatGPT is biased against conservatives" allegation without going all into a grievance mongering paranoid "this is yet another example of how the Evil Forces of Wokeness have assembled to marginalize conservatives" narrative, as some other right-leaning outlets have done.

    Collapse
    1. Joe1776

      Yeah I mean I find it annoying that they have to put this into everything and I am glad he does note that it is happening. But there is no reason to lose sleep over it. I think everyone knows anything that silicon valley touches will involve some level of leftist propaganda.

      Collapse
  5. Avatar photo
    Linda Beggs

    The brave, new world of AI gives me the creeps.

    We all have experienced 'coincidences' where the most logical explanation is the cell phone is listening--even when we're not actively using it.

    I read (maybe in TD comments section) about someone's printer (which they bought, not leased) being remotely disabled by the manufacturer because they exceeded the number of pages allowed under their ink plan. An example of forces 'out there' limiting our choices.

    I want to interface with actual humans, not machines. At least not until they get to the level of C3PO (which then introduces a host of moral questions about ownership.)

    All this makes me feel like some kind of paranoid old crank. But there you have it.

    Collapse
    1. Joe1776

      I just find it frustrating when companies waste my time with these robo call centers telling me information I am not interested in.

      Next time I talk to a person at Walgreens before I hang up I want to say "Did you know My office hours are 8:30-4:30? And my office is located on the northwest corner of veterans highway and route nine. And by the way I take a meal break between 12:15 and 1 pm every day. "

      Edit: so yes I have "out cranked" you.

      Collapse
  6. Avatar photo
    EricRemcon@gmail

    Thanks for the piece, Alec. I agree AI is here to stay and we will need to adapt. But (ala Scott Lincicome) there is no need to panic. I would guess customer support and web searching is where we will see the impact first. I also strongly agree with the last paragraph. Many are concerned about losing their jobs or nuclear missile silos under AI control or smart attack drones/robots with shoot/kill programming. The more mundane worry is that the product of AI will have the flavor (bias) of the non-sense that is prevalent online nowadays. We certainly don’t need more of that.

    I do have one jab though. The survey quote that was included with no context bothered me.

    “By the end of 2022, 50 percent of respondents to a McKinsey and Company survey said they were using AI in their business in some form.”

    “some form” is doing a lot of lifting there. I would guess (but don’t know because I didn’t look for the survey) the majority of the respondents had those silly voice recognition call systems or automated e-mail reply algorithms in mind. Those schemes are very tangentially related to the AI the article is about. The quote inclusion struck me as something you would see in a USA Today article.

    Collapse
    1. Avatar photo
      Aylene Wright

      "Customer service chat" functions on the vast majority of corporate sites are obviously NOT being staffed by actual "meat" people. Some are even quite upfront about that, although not always upfront as to how to actually reach a real person, not a bot.

      I recall a really hilarious article some years ago on how the writer got a customer service chatbot to admit it wasn't a real person.

      Collapse
      1. Avatar photo
        EricRemcon@gmail

        You’re right, I should have included those annoying “what can I sell you” or “let me help you navigate our website, you ignorant human” chat bots. I don’t know their inner workings, but I doubt they are anywhere near as sophisticated as what is being referred to with the term AI.

        Collapse
  7. Avatar photo
    Chris L

    It will be impossible to get everyone on board with this, if only because too many people absolutely refused to be told things that are even marginally inconvenient to them. For example, when BCS computer rankings first came out about 20y ago, football fanboys went absolutely apoplectic when they were informed that their favorite teams weren't ranked all that great, let alone when this was confirmed by their teams' subsequent beatings at the hands of teams previously marginalized ("noo! the computers must be wrong!!!"). How did they solve this conundrum? Computer rankings now account for only one third of the BCS formula, and even that third has large human elements.

    Collapse
    1. Avatar photo
      Dusty

      The BCS is dead. The CFP is entirely decided by a committee of humans.

      Though, this would have been a great year to have the BCS.

      Collapse
  8. GregS

    "Will these artificial intelligences ever reach the point where they can distinguish truth from fiction or process information without tacitly adopting a particular point of view?"

    No.
    We don't know how to get natural, human, intelligences to do that reliably. Why would we be able develop artificial ones that can?
    Furthermore, even if we developed an artificial intelligence that was unbiased, how would we objectively be sure of it? Your examples of bias in the piece are perfect examples - some people would agree they are "biases" and others would say they are just ,"values." A truly unbiased AI is an amoral AI - do we want that, and would we recognize it as intelligent or just monstrous?

    Collapse