2023 will likely go down as a year of AI renaissance, sparked by advancements in generative AI and a hype cycle to boot. It wasn’t all puffery though, as we’ve really made some significant strides forward this year. Here are some highlights…

2023: The Year of AI Renaissance

1. Transformers Come Of Age

AI has come a long way since the 50s, but it’s transformers that really changed the game. These were Google’s big idea back in 2017, and they’ve totally reshaped how we think about AI that can generate stuff (hence, “Generative AI“). Think of transformers as super-smart digital helpers that are great with words. They can spin up stories, make art, and crack tough problems. By 2023, they’re leading the AI pack, turning simple text into amazing pictures and deep insights. They’re really good at picking apart language and spotting connections that we might miss.

But it’s not all smooth sailing. Sometimes they get things wrong or end up using a lot of power, so we’ve got to think about keeping things green too. Just this year, we’ve had a flurry of developments: Microsoft put a ton of money into OpenAI, we saw new Large Language Model (LLM) enabled Chatbots to compete with ChatGPT, like Google Bard and Bing Chat, and there was the launch of the super-advanced GPT-4. Then there were interesting releases like AutoGPT, Claude 2 from Anthropic, and Llama 2. Plus, ChatGPT got some awesome upgrades late in the year (like the ability to see, hear and speak), Elon Musk came out with Grok (a quirky chatbot inspired by A Hitchhiker’s Guide to the Galaxy), and Google updated Bard with Gemini.

I started my own experiments with Generative AI early in March and blogged about it as part of the 30 Day AI Challenge, followed by a series of posts under the AI Summer 2023 heading. It’s been a wild ride this year…and we’re just getting started.

PS – did you know that there are over a dozen AI chatbots out there, powered by large language models? Here’s a fun experiment with 12 of these AI chatbots, including a ranking of their capabilities.

2. New Creative Boundaries

Generative AI has really shaken up the creative world, both in amazing and challenging ways. Think of it like a double-edged sword for creativity. On the plus side, it’s opened doors for artists and creators, making things like image editing, video creation, and even 3D modeling a breeze. Just look at what’s happened this year: In January, ElevenLabs brought out the best text-to-speech tool (yet), and we got Instruct Pix2Pix for tweaking images. February saw AI art getting even cooler with Control Nets, that offered more image consistency. March was huge with MidJourney making more realistic AI images, Adobe jumping into AI art with Firefly, and Runway’s first video generation tool.

Artists like Refik Anadol have embraced AI to push the limits on what’s possible artistically.

But it’s not all perfect. There’s worry about jobs in creative fields and the originality of art. Still, the progress keeps coming. Meta’s tool in April for editing images and videos, Wonder Dynamics’ tech for swapping real actors with CGI, and OpenAI’s Shape for turning text into 3D models are just a few examples. Runway’s Gen 2 in June improved video generation even more. Then came DALL-E 3 in October, making AI images stick closer to what you ask for (and right within ChatGPT), and Adobe’s cool text-to-vector tool in Illustrator.

Above: A comparison of Mid Journey v4 (Jan 2023) output with v6 Alpha (Dec 2023).

By the end of the year, we’ve seen even more: Gen 2’s Motion Brush for videos, Latent Consistency Models for real-time editing, and advanced text-to-video with P 1.0. And December? MidJourney v6 Alpha, the most advanced text-to-image model yet. It’s clear that Generative AI is changing the game in creativity, for better or worse.

3. A New Arms Race

Generative AI has kicked off a real competition among big tech companies and countries, and it’s a mix of exciting progress and some tough challenges. Think of this as the Oppenheimer moment…for AI. Back in January, Microsoft put a whopping $10 billion into OpenAI, really showing they mean business by mixing this AI into their products. Then in February, Google jumped into the chatbot game with Bard, but it didn’t quite match up to ChatGPT, and Microsoft was quick to launch Bing Chat using GPT4, making a big splash.

By May, Google was pushing AI into all sorts of products, ramping up the competition. July saw Anthropic trying to outdo ChatGPT with Claude 2, and Meta went a different route by making their Llama 2 model open-source, shaking things up. Amazon committed up to $4 billion in Anthropic, with the commitment to being their primary cloud provider. Fast forward to November, and there’s drama at OpenAI with Sam Altman’s firing and rehiring, showing there’s tension even within these tech giants.

In December, Google tried to wow everyone with Gemini, their huge AI model. They called it a multimodal marvel, but got some heat for not being fully upfront about a demo video (seen above), raising flags about transparency and maybe overselling what their AI can do. This whole race is pushing the boundaries of AI, which is great, but it’s also a reminder that we’ve got to keep things honest, plus keep a firm eye on safety…and not just get caught up in the hype.

Bonus: here are just some of the safety and security risks posed by AI (watch the playlist of YouTube shorts below). More: The AI Intimacy Trap | AI Characters

Yet another challenge (and arms race) exists on the hardware side, with the majority of the world dependent on TSMC (Taiwan Semiconductor Manufacturing Company Limited) for the supply of cutting-edge chips that are powering the AI revolution. Here’s a great video that explains the challenge:

4. Amplified Conversation

AI, especially Generative AI, isn’t just tech talk anymore—it’s a hot topic for everyone, stirring up big conversations about where our society’s headed. We’re talking utopian dreams and dystopian fears, and the real deal is probably somewhere in between. This year, AI’s societal impact was front and center.

In May, Jeffrey Hinton (the Godfather of AI) left Google, worried about AI’s future safety and security. That’s a big deal, right? Then there was a Senate hearing on AI regulation, showing the government’s taking this seriously. Fast forward to November, and the UK’s AI Safety Summit brought together big names from all over, like the US and China, to talk about keeping AI safe and sound.

But the real game-changer was in December with the EU’s Artificial Intelligence Act. This is major—it sets rules for AI, deciding what’s okay and what’s not, and it’s aiming to be a global standard. They’re talking about checking AI systems before they hit the EU market, though some folks are saying they could be tougher on the high-risk stuff.

So yeah, AI’s definitely in the spotlight now, and it’s sparking all sorts of talks about its role in our future. It’s like we’re all trying to find that middle ground between the good and the bad of AI.

If you’re interested in hearing from the world’s AI thought-leaders, here’s a curated playlist of the best AI TED Talks from 2023:

5. AI Adoption: Different Industries, Different Speeds

Alright, so when it comes to AI, especially the Generative kind, it’s like a wildfire spreading through different industries. Some are jumping in faster than others. Tech companies, for sure, are leading the charge. They’re all about using AI for stuff like chatbots and image generation. Then you’ve got creative industries, like graphic design and music, where AI’s helping create new kinds of art and tunes. Even more traditional sectors like healthcare and finance are getting in on the action, using AI for things like analyzing data and improving customer service.

The fast movers are seeing some great benefits. AI’s making things more efficient, sparking new ideas, and even cutting costs. But it’s not all smooth sailing. There are worries about jobs – like, will AI replace people? Plus, there’s the whole can of worms about ethics and privacy, especially when AI’s deciding stuff that affects people’s lives.

So, it’s a mixed bag. Industries are keen to adopt AI for the good stuff, but they’re also treading carefully because of the challenges. It’s all about finding the right balance, making the most of AI while keeping an eye on the potential downsides.

The travel space has been slow on the take, as usual, largely owing to legacy tech stacks, slower investment cycles and worries about safety and privacy. There were a few contenders (mainly OTAs) like Expedia and Kayak that hopped on the Gen AI bandwagon by launching ChatGPT plugins early in the year, but largely hotel and airline companies are still treading water and working on more robust use cases behind the scenes.

2024: Five AI Predictions

Next year promises to be a roller-coaster year for AI – here are five predictions:

1. Transformers may not be the only game in town: The capabilities of LLMs will be more closely scrutinized, with a growing awareness of their limitations and the diminishing returns on increasing model sizes. This could lead to a focus on more efficient, specialized models, possibly multimodal, offering similar effectiveness with easier and less costly updates​. Alternatives to transformer architectures, like the Mamba model, are also poised for adoption, potentially offering more efficient AI processing. Even OpenAI might introduce a smaller but powerful language model, signaling a shift from the dominant transformer models.
2. Creative boundaries will continue being pushed…as will the law: Generative AI will advance in creative fields, but legal challenges will rise (and some likely be addressed), especially around copyright issues. AI applications like agent-based models and generative multimedia (video and music) will evolve from experimental stages to more practical and sophisticated uses.
3. The Arms Race will heat up, with developments coming at a dizzying pace: Closed models like OpenAI’s GPT4 (and soon GPT5) will likely continue to outperform open-source models (like Meta’s Llama-2) given the level of investment required. Companies like Nvidia may expand into cloud services, directly competing with current cloud providers. OpenAI will continue to evolve into a product-centric company, possibly releasing new AI tools and applications, including an “app” store for AI. Apple will jump into the fray with a more serious focus on practical AI use cases, leveraging their device and user ecosystem to offer “agent” type AI capabilities (think a ‘smarter’ Siri) with an emphasis on privacy and security. On the hardware side, the race to reduce reliance and risks on the semiconductor supply chain will heat up, with the US, China and other key players making strategic moves.
4. Some of the conversation will turn into action, especially in terms of regulation: Anticipate new regulatory actions and an increase in legal cases involving AI. The industry will likely witness an early adoption of AI compliance standards, possibly influenced by regulations like the EU’s AI Act. AI will likely play a significant role in the 2024 US (and potentially other major powers like India, the EU and the UK) elections, with concerns over its use in spreading misinformation, manipulating narratives, and influencing voter perceptions.
5. AI Adoption will accelerate, even in the travel industry: Broad adoption of AI across industries will continue, with significant investments and strategic moves (including some of the politics we’ve seen in 2023) from major players like Microsoft, Google, and Apple. Some major companies will likely create new C-Suite positions dedicated to AI to provide the right level of focus on adoption and safety. The travel industry, along with others, is likely to increasingly integrate AI for backend efficiency (like simplifying data entry, knowledge base access in natural language, etc.) and some low-risk customer facing innovation (e.g. curated chatbots, trip inspiration and more).

These predictions indicate a year of dynamic growth and challenges in AI, with significant transformations in technology, regulation, and industry adoption. If you’re interested in looking even further, check out this post I did on what AI will look like in 5 years?.

Learn About AI (Or Get Left Behind)

AI’s becoming a big deal, and it’s changing things faster than ever. Now, it might not snatch away your job, especially if you’re in a white-collar role, but get this: the future’s going to be about who can use AI the best. So, it’s super important to jump in and learn all you can. If you’re keen to get the lowdown on AI and sort out what’s real from just hype, especially about Generative AI, check out this video I made. I’ve been sharing it with my work pals for half a year and thought it’d be cool to put it online for everyone. You can watch it right here, or head over to YouTube for a full-screen experience on your device or TV.