Could this arcade solve the world’s problems through AI?
Around the office, we found a big red button from an old ad campaign, waiting to resume its purpose and start being pushed again. So what’s the most amazing thing this button could trigger?
If AI is as amazing as my Twitter feed is leading me to believe, it should be able to fix the world’s biggest problems with the click of a button… right?
Well, let’s see what this technology can do. We had been testing the AI chatbot GPT-3, and using the UN’s list of the world’s most pressing issues, we prompted the AI to find a solution using emerging technologies. The results were surprisingly… decent. Not world-changing, but some serious sparks; and the more we worked on the prompt, the better the answers became.
We then designed a retro game experience with bespoke music and lighting. We installed an old screen, 3D-printed our logo and built an arcade-style casing box. Lastly, we asked AI to come up with a name, and “Eureka! The Idea Arcade” was born.
We then placed the machine in the middle of the office:
So please, go ahead and press it yourself. Not in Droga5 Dublin yourself right now? No problem: play the arcade’s digital double → EurekaArcade.com
Is AI going to steal my job?
We appreciate that not everyone wants to press the button. Running alongside the surge in AI excitement is a steady stream of confusion and hesitance to engage. It’s all a bit techy and confusing, with an added whiff of job insecurity in the air.
We fear what we don’t understand, and yet, there’s no turning back. So, let’s get started with ChatGPT3 and start by giving the AI a chance to defend itself:
That all sounds reasonable, but hesitance is fair; we need to take a watchful eye and human approach. I for one don’t think AI is going to send you packing. AI isn’t going to steal your job, but the person who knows how to use it will. So click the button already. It’s jaw-droppingly powerful but also riddled with issues, so it takes a human to harness this new tool and weed out the robo-drivel.
Pros
Good mistakes — It can draw on unlikely relationships that we humans wouldn’t even let hit the page. This can, of course, lead to a bunch of bad ideas, but sometimes an unlikely relationship between concepts can develop and spark new thinking.
Donkey Work — My dream is for AI to do my timesheets. Already AI is helping people with monotonous tasks. I’ve used ChatGPT3 to reformat Excel sheets, and it may have even helped me write a few performance reviews… :)
Writer’s Block — Sometimes the hardest part is getting words on the page, even if they’re sh*t ones. Say you have to write a press release or a new scope-of-work agreement; ChatGPT understands the format and context you’re writing in, and can draft a framework to suit, enabling you to go from a writer to an editor, and gain a new lease of life.
Code monkey — ChatGTP does a phenomenal job understanding code. It can’t solve or come up with complex solutions, but it will clean up code, or make it more efficient. This is shifting the effort and value of coding (and all kinds of writing), allowing more time to think of the big picture, while quickly fixing the little stuff.
Dissertation — I was explaining GPT-3 to a college student recently, prompted ChatGTP with their thesis, and got back their essay point by point. Is the prompt the new thesis statement? If you’ve already done the thinking, this can quickly articulate it. This could get interesting for the education system.
Cons
Out of date — AI Chat (at this moment) isn’t like Google,trained up to the minute of the internet for facts like whether that restaurant is still open, or who won the game last night. It’s currently drawing from data from 2021, but it does continue to get better.
Facts — Openai is very clear with their warning when using it — there are lots of mistakes in what it gives you, especially when it comes to facts and figures. You should always double check with other sources.
Creativity — It’s only as smart as we’ve taught it to be, so original ideas aren’t its thing. But it can be a helpful buddy to bounce things off of. Eureka! is our way of showcasing this.
Bias and racism — When testing the results on the arcade, we found that GTP-3 didn’t realize if xenophobia was good or bad. For example, it started coming up with results for how extremist groups could connect using VR. So you have to guide the prompt to get the stance right. This is what happens when you train something from things people say on the internet.
So now, I’ve asked you three times. What are you waiting for? Smash it.