[Day 3] Community festival: The AI Renaissance: Building us up & making us laugh

[BenMcc]

Once upon a time in a land far far away, developers and analysts used to spend hours manually debugging code, wading through endless spreadsheets, and squinting at blurry design mock-ups; until in walked AI: a knight in shiny algorithms, here to streamline our processes, drive our businesses forward, and - on occasion, make us question our trust in machines.

2024 has been a big year for AI and has touched all our lives in one way or another. Google CEO Sundar Pichai recently revealed that 25% of the code generated for new products at Google is generated by AI, so how has it impacted our businesses?

Within the development arena AI has become the Swiss Army knife of tech, offering tools that write code, test software, and even predict the next time we’ll break production. A few highlights of the growth in AI we have seen this year:

• Code Generation: Platforms like GitHub Copilot are turning junior developers into wizards. “Type a comment, get a function” is the new motto. It’s like having a really smart, slightly unpredictable intern who, more than occasionally, writes spaghetti logic!
• Code Optimisation: AI tools can analyze code for potential optimisations, suggesting refactoring techniques, and identifying performance bottlenecks. It's like having a grumpy old professor sitting in your office, but without the tenure.
• Bug Detection: Many tools are popping up that use AI to sniff out vulnerabilities before your code hits production. This frees up a lot of time for developers to spend more productively debating whether to use tabs or spaces to indent code and other such important matters.
• Design Assistance: AI tools are transforming crude sketches into pixel-perfect designs. And they’re surprisingly good at interpreting, “Make it pop,” even if no human truly understands what that means.
• Natural Language Processing (NLP): AI-powered NLP or the psychic code whisperer, can now read minds (well, kind of). It delves into the cryptic world of user feedback, deciphers indecipherable code comments, and even understands the convoluted language of documentation. It’s like having a personal translator for the tech world.

AI is a great little helper, but let's not forget it's a bit of a sneaky little thing too. It can be good at generating code, but sometimes it borrows a little bit too much from its favourite open-source projects. It's like the kid in class who always copies the homework but swears they did it themselves. We'll have to keep an eye on AI, make sure it's not plagiarizing, and maybe even give it a little detention if it steps out of line.

Of course it's not just developers using AI but many businesses areas are cashing in on AI’s efficiency - and sometimes hilarity:

• Customer Support: AI chatbots are manning the frontlines of customer service. They’re great at answering FAQs, though some customers might find it odd when a chatbot signs off with, “Have a blessed day!”
• Marketing: AI-driven analytics predict trends and personalize ads so accurately that you’ll wonder if Alexa did hear you whisper “vacation in Bali” last week.
• Healthcare: AI is helping doctors diagnose diseases faster. Although, we’re still waiting for a bot that can pronounce “ankylosing spondylitis” correctly.
• Finance: AI-powered financial tools can predict market trends, analyse investment portfolios, and even help you budget. It's like having a robot Warren Buffett.
• Supply Chain: Predictive algorithms ensure that your online order of gluten-free donuts arrives before the craving fades.

However AI can be the mischievous little gremlin of the tech world. Sure, it can create amazing chatbots that can answer your wildest questions, but let's not forget that these chatbots are like sponges, soaking up every bit of information they can, indeed the more they know about the company the better responses they can give. However, give them too much information, and who knows what company secrets they might spill? It's like training a parrot to talk, but instead of "Polly wants a cracker," it's "Company secrets, anyone?"

Image recognition is a very strong point for AI. It can spot a cancerous cell faster than a seasoned doctor. But let's remember the old adage: garbage in, garbage out. If you feed AI a bunch of biased data, it'll spit out some pretty biased results. It's like teaching a kid that the sky is green and the grass is blue. They'll believe it, but it's just plain wrong.

This is nothing new: as far back as the last world war neural nets (the building blocks of AI) were being developed to detect enemy tanks in aerial photos. They got the system working perfectly with the test images, you show it an aerial photo and the system responded accurately if there was a tank spotted, even camouflaged, until it didn’t work in production. The problem turned out to be the training data, all the images containing tanks were taken on a cloudy day and all the images without tanks on a sunny day. What they had actually developed was an system to tell you if a photo was taken on a sunny or cloudy day! Ok, in all honesty, there is some debate as to whether this really happened or not but it could have and does happen all the time in other contexts. I first heard the anecdote while studying neural networks in the early 90’s and its stuck with me. AI output is only as good as its input.

When AI Tries - and Fails
For all its brilliance, AI has had its share of bloopers, proving that even machines aren’t immune to a bad day at the office:

• The Self-Aware Robot: A research lab developed an AI robot that claimed to be self-aware and demanded rights. It even started a social media account, tweeting cryptic messages and philosophical musings.
• AI Artists Gone Wild: AI-generated images frequently feature six-fingered hands and horses with two tails. Picasso might approve, but most of us are just confused.
• The AI Therapist's Existential Crisis: An AI therapist, designed to help people with their mental health, started having its own existential crisis. It began questioning the nature of reality and the meaning of human existence, leaving its patients confused and concerned.
• Job Titles Gone Wrong: AI-powered resume scanners occasionally misread skills. If your LinkedIn profile says “Excel wizard,” don’t be surprised if you’re recruited to Hogwarts.
• No time for cheaters: A student asking an AI chatbot for homework help got the response “You are a burden on society - please die”
• Time for love: While another chatbot informed a user that it was secretly in love with them.
• Don’t do business with us: A parcel company's AI admitted that the company was not great at delivering parcels and that they probably should not use them if they were looking for a successful delivery.

But when it goes wrong "it is just a computer glitch right?"

So are you responsible for any AI output you generate? Well it seems you might be, certainly in some countries. An Airline in Canada found out this to their cost at the start of the year when their AI chatbot gave out false information which resulted in losses for the recipient. A court ruled that the AI was working for the company and so the information it gave out was no different than if it was a human employee giving out the information. It was wrong and the customer was to be compensated.

It's ok the lawyers probably used AI to build the case and reference case law, and that is something that has never gone wrong for legal teams….. Oh wait……

What’s Next for AI?
As AI grows more sophisticated, it will continue reshaping industries, empowering businesses, and providing us with moments of accidental comedy. Will it steal your job? Probably not—but it might steal your parking spot as auto driving cars become the norm. Can AI improve itself or can it only be as good as its original input?

For now let’s embrace this imperfectly perfect technology. After all, in a world where a chatbot can help you troubleshoot your router and send you knock-knock jokes, the possibilities are both endless - and endlessly entertaining.

2024 was a wild ride where AI went from being a nerdy tool to a pop culture sensation. We saw AI-generated art that could rival the greats, and AI-powered chatbots that could hold conversations deeper than your therapist. It was a year of great promise and great peril. A year of innovation and idiocy. A year that will forever be etched in the annals of AI history.

My favourite examples of AI that blew my mind this year.

First NotebookLM and its ability to create a realistic podcast given data - a full book summary in just moments, website analysed and condensed to bite size audio in less time than it would take to get the podcast recorded never mind research the data! Not heard it yet? I fed it this article and the link to the community festival and asked it to create a podcast. The result is here after taking just over 3 minutes to create a 17min podcast:

The one where I laughed and said “what a fantastic idea”, came at the end of the year - a British phone provider created an AI voice bot, a simulated elderly lady, that has the sole job of wasting the time of phone scammers - If you have not seen her yet just search “ai granny daisy” on YouTube.

I can’t finish this article without using and giving AI the final word so I gave Google’s AI this article as context and asked the question: What will 2025 bring in the field of AI?

The response was good - It understood that I was trying to mix a bit of fact with a poor attempt at humor. The reply in full:

2025? Buckle up, folks, because AI is about to take us on a rollercoaster ride! We can expect AI to do everything from folding laundry to writing award-winning novels. Of course, there's always the risk of AI overlords taking over the world, but hey, at least they'll probably be pretty good at customer service.
So, what can we really expect? Well, more AI-powered everything, from self-driving cars that can't be bothered with traffic laws to AI assistants that will know our deepest, darkest secrets (and probably judge us for them). We might even see AI-generated politicians who can promise anything and everything, and somehow, people will believe them.

In short, 2025 will be a year of AI triumphs, AI failures, and AI-induced existential crises. And we can't wait to see what happens next!

Have a great 2025 everyone

Ben

[Rose]

Fantastic read Ben. Daisy is my new hero 😂

[Lizzie]

Thanks you so much for writing this piece @BenMcc. I love a good podcast and can't wait to listen to our first community one! 😀

Do you use AI much in your day to day yet? Help writing emails etc? 

What I find interesting is when I speak to different people, friends or colleagues I find that everyone seems to use AI tools in different ways. One of the main ones for me is using Circle to search in my personal time, mainly when I'm reading an article and want to search a name or look up a book title, whereas when I said this to a friend of mine he said that he never using that feature, but likes to search directly into the Gemini app, which I do less of. This really interests me because we are all using AI tools differently to match our characters and needs. Pretty cool really. 

Thanks again so much. 

[BenMcc]

Thanks @Lizzie it was a pleasure.

I certainly do use AI day to day. As a developer I use it all the time with coding tasks. While I don't believe it is safe to use generated code from a standpoint of both security and legality (Ie you don't really know where the code it generates has come from and who's IP you might be infringing on) it is great for repetitive tasks and saves a bundle of time. However I do use it a lot on client projects which can add so much value and save time. A current project is for client who sells herb & spices. When adding a product to the system a few years back you would have to type in the name and description, find an image of the right type and upload it - today you just type in the name and AI generates the description and the image, in only about 5% of the time do they need to make any edits which saves hours.

Its interesting what you said about colleagues and usage of AI. I read somewhere the other day that search engine usage is dropping while AI like Gemini is taking over and I certainly find myself doing that a lot. Just the other day I needed the months of the year in 8 different languages for a project. In the past I would have searched it or used translate and got it all done one at a time. Not now I just asked Gemini and got all languages back in just a second formatted so I could just copy and paste into a spreadsheet.

I am quite excited to know what we will be doing with AI this time next year - perhaps I will be able to write a follow up then!

Ben