At Least This AI is Honest about Its Inadequacies

Prolific but bland Bard struggles with presentation and podcast.

GUEST COLUMN | by Angelo Fernando

GALIH WARSITA

I began testing ‘Bard,’ Google’s AI, last week. I have been holding ChatGPT’s feet to the fire, too; more on this later. Before I got started with using Bard—for which I had to apply and wait for an invitation that came in a few days—Google hit me with some cautionary notes. Perhaps to lower my expectations. Here’s the first: “I’m Bard, your creative and helpful collaborator. I have limitations and won’t always get it right, but your feedback will help me improve.”

‘Before I got started with using Bard—for which I had to apply and wait for an invitation that came in a few days—Google hit me with some cautionary notes.’

It was followed by this pop-up about expecting the ‘inaccurate’ and ‘inappropriate.’

So here’s what I set out to test. I’m a teacher. I didn’t want to ask the AI to write an essay, or comment on a popular topic. I wanted to probe deeper into its machine mind. To see if Bard could come up with content that would fit a presentation. To see if it understood what makes a good set of slides. And then I would follow up in an unexpected way.

My prompt went like this:

Create five slides for a presentation titled, ‘The Future is Now.’ (I teach a computer class and recently had my students in 7th grade work on a similar presentation. I also teach a High school class in writing and publishing. Could Bard match their work?)

Here’s what its AI came up with: (Note: Bard did specify slide-by-slide points. For your convenience, I created a deck in 5 Google slides and pasted its content.)

My Take

1. Bard made some good points, especially since it touched on robotics and 3D printing, and the expected reference to itself, AI.

2. It was way too wordy for slides. Sentences such as “Despite the challenges, the future also holds great opportunities” are phrased in a way that might work in an essay, not in a snappy presentation..

What I liked:

(a) It was free of tech jargon.

(b) The sections on ‘Trends’ and one on ‘Challenges and Opportunities.’

What I hated:

(a) It’s boring. The first two sentences were very bland — basically stating the obvious in a boring way. I would mark down a student who states the obvious in this fashion.

(b) It’s blather. It doesn’t know how to write or tighten bullet points. In other words, it didn’t get the memo we all got many decades ago: Slides should have bullet points not sentences. Much of it could be speaker notes.

(c) It’s bloodless. AI-generated language predictably exhibits the passion of a boilerplate. Who blurts out things like “a more prosperous, equitable, and sustainable world.” No one you and I know.

But Wait! There’s More

I wasn’t ready to let Bard go.

I next asked Bard to “Turn this into a 5 minute podcast script. I asked it to work with a host named ‘Equinox Chatterbard,’ and a guest named ‘Radium Kurzwild.’ Also I asked it to “suggest a music track.” I was expecting it to regret its boilerplate nonsense and breathe some life into dialogue.

I waited with bated breath. Here’s what it came back with in under a minute:

Title: The Future is Now

Host: Equinox Chatterbard

Guest: Radium Kurzwild

Music: “The Future” by Daft Punk

Equinox Chatterbard: Welcome to The Future is Now, a podcast about the future of technology and society. I’m your host, Equinox Chatterbard. And today, I’m joined by Radium Kurzwild, a futurist and author of the book “The Future is Now: A Guide to the Technologies That Will Shape Our World.” Radium, thanks for joining me today.

Radium Kurzwild: It’s my pleasure.

Equinox Chatterbard: So, let’s get started. What do you think is the biggest trend that’s shaping the future?

Radium Kurzwild: That’s a great question. I think one of the biggest trends that’s shaping the future is the rise of artificial intelligence. AI is already having a major impact on our lives, and it’s only going to become more powerful and sophisticated in the years to come. AI has the potential to revolutionize many industries, from healthcare to transportation. It also has the potential to create new jobs and opportunities.

Equinox Chatterbard: What are some of the potential benefits of AI?

Radium Kurzwild: AI has the potential to improve our lives in many ways. For example, AI can be used to diagnose diseases, develop new drugs, and create personalized learning experiences. AI can also be used to automate tasks, which could free up our time to do other things.

Equinox Chatterbard: What are some of the potential risks of AI?

Radium Kurzwild: AI also poses some risks. For example, AI could be used to create autonomous weapons that could kill without human intervention. AI could also be used to create surveillance systems that could track our every move. It’s important to be aware of these risks and to take steps to mitigate them.

Equinox Chatterbard: What do you think the future of AI looks like?

Radium Kurzwild: I think the future of AI is very bright. AI is going to continue to develop and become more powerful. It’s going to have a major impact on our lives, both positive and negative. It’s important to be aware of these changes and to prepare for them.

Equinox Chatterbard: Thanks for your insights, Radium. That’s all the time we have for today. But be sure to tune in next time for more on the future of technology and society.

Music: “The Future” by Daft Punk

So What Do I Think?

Bard did a half decent job with my two fictitious hosts. However the script didn’t capture or hint at the personalities. Maybe I’m biased – having a podcast. It’s typical, though not required, for a host to say something provocative about the moment host and guest meet, or why he/she chose this guest. Bard only let its host, Equinox, refer to the book, but not what about the book made it interesting. Equinox jumped straight to the Q&A.

I was not surprised. AI being ‘raised’ on content and databases rather than idiosyncrasies of people, or unpredictable bursts of imagination, doesn’t know this—yet.

My Takeaways

The podcast guest, Radium, sounded like a PR mouthpiece for AI. No surprises, right? It spouted a paragraph of cliches. (“AI is already having a major impact on our lives, and it’s only going to become more powerful and sophisticated in the years to come.”) Radium also steered away from making personalized (even controversial) observations. That’s what you would have expected of a guest in a podcast, right?

The podcast host, Equinox rephrased the ‘on the one hand… on the other hand” point from slide #4 above, making it a dull, safe response. (What’s referred to as ‘both-sidesism.’) His response when asked what the future may turn out to be was what I would have expected as the closer of a 4th grader’s essay: “I think the future of AI is very bright.”

Music: Not a bad choice. Daft Punk did have other tracks worth considering, such as “Human After All.” Perhaps Bard isn’t clued into these lyrics, which is why it chose ‘The Future.’

My Final Analysis

Bard lived up to the low expectations Google set for me. Should we educators be concerned about this or ChatGPT or other similar apps that are in the new AI arms race? I am not perturbed. Many of the alarm bells that have gone off, thanks to the media doing barely researched stories into ChatGPT, are what I expected. It reminds me of our Wikipedia moment many decades ago, when people thought it was a pox on society.

AI, like search engines, or other algorithmic apps we now use will be a part of our lives, so we better get to know and question how they work. It’s a critical edtech juncture, where the ed and the tech will meld together faster than we might keep up.

‘AI, like search engines, or other algorithmic apps we now use will be a part of our lives, so we better get to know and question how they work.’

I recently conducted tests using ChatGPT, giving it some creative assignments. This included asking the AI to write a song in the vein of Bob Dylan. It was a total bust. These modern day scribes (or muses) are still in their infancy. They will get better. But for now they can’t beat the creativity of my seventh graders. Which could be more devastating to some than this fancy crop of AI failing the Turing Test.

—

Angelo Fernando is a teacher at Benjamin Franklin High School in Queen Creek, Arizona. Connect with him on LinkedIn.