As we bid farewell to 2023, let's take a trip down memory lane and revisit the most impactful events in the world of AI. From groundbreaking innovations to juicy drama, this year has certainly been one for the books. So, buckle up and get ready for a wild ride!

ChatGPT-4: The Bigger, Badder Brother

One of the most significant events in 2023 was the release of ChatGPT-4. This powerful language model, the successor to the wildly popular ChatGPT-3, brought generative AI to the mainstream. With its ability to understand and generate human-like text, ChatGPT-4 made AI accessible to the masses, sparking a renewed interest in the technology. Some key takeaways from ChatGPT-4's release include:

Accesses image inputs and produces image outputs, even solving CAPTCHA and ReCAPTCHA codes
More functional, responsive, and safer than GPT-3 or GPT-3.5
Passes a simulated bar exam with a score around the top 10% of test takers, compared to GPT-3.5's score in the bottom 10%

I personally was using ChatGPT-4 since it became available to the public. Unfortunately it was only available with a subscription, but it was one of the subscriptions I was happy to pay, because it really made my life much more efficient than I had imagined. I would use it to give me actionable feedback and ask me questions to make me think of aspects I might not have thought of. Basically it was a colleague that wouldn’t perceive me in any way, but just be helpful.

Claude 2.1: The Competition Heats Up

Not to be outdone, Anthropic unveiled Claude 2.1, its own cutting-edge language model update. This release showed that the AI landscape is growing more competitive by the day, with companies vying to create the most advanced and versatile language models. Some highlights of Claude 2.1 include:

A 200,000 token context window, allowing the AI to process more text and provide better summaries, question answering, trend forecasting, and other insights
A two-times reduction in false or hallucinated statements, making the AI more accurate and reliable
System prompts, which allow users to provide customized instructions and context to improve the AI's performance
Tool use, a new beta feature that enables Claude 2.1 to interact with developer-defined APIs and integrate with users' tech stacks

I’m currently am experimenting with it and it seems like it can produce much longer responses, which I don’t necessarily like, but I don’t have my final verdict yet.

Google Gemini: Unleashing the Future of AI

On top of all that Google released Gemini, its long-awaited answer to ChatGPT. It’s currently not available in Ireland, so I haven’t been able to fully test it without the use of a third-party application, but Google did make a big deal about it, quickly fizzling out once people read the fine print, which made it quite similar to ChatGPT-4. Having said that, when I tried it, the answers provided were super fast, but they lacked correctness. You can say that I’m really excited to see where Google takes it. Here's what you need to know about this exciting new development:

Unlike other AI models that focus on text, Gemini is "natively multimodal," meaning it can understand images, video, and audio, in addition to text
Gemini is designed to be efficient and fast, thanks to Google's custom Tensor Processing Units (TPUs)
Gemini is being integrated into various Google products, including search, ads, and Chrome, with more to come
Google sees coding as a killer app for Gemini, which uses a new code-generating system called AlphaCode 2 that performs better than 85% of the time

Dall-e 3 vs. Midjourney v6: The Battle of the AI Artists

In the world of AI-generated art, 2023 saw a fierce competition between OpenAI’s DALL-E 3 and Midjourney v6. Each platform showcased its unique capabilities, pushing the boundaries of what's possible with AI-generated visuals. Here's how they stack up:

DALL-E 3:

Produces fine detailed images
Has a more naïve approach to image making
Offers a standalone website version and API
Provides more restrictive censorship

Midjourney v6:

Excels in photo realism and atmospheric design
Offers more control and parameter changes
Has a more supportive and communicative community
Has limitations on image generation

On top of all that, both support adding text to images that no longer generates complete gibberish, which has been an issue for image generators for a very long time with very few exceptions.

Meta also released an AI tool, but I have not been able to get access to it at all due to it being restricted to specific geos, but I really enjoyed John’s comparison of the most popular AI image generators, so be sure to check it out here.

The Drama Surrounding OpenAI

No recap of 2023 would be complete without mentioning the drama surrounding OpenAI. From internal conflicts to public controversies, the company faced its fair share of challenges. I won’t go into too many details, but some key moments include:

The departure of key personnel, then joining Microsoft, then getting hired again by OpenAI
Debates over the use of AI in content generation
The ongoing struggle to balance innovation with responsibility

Predictions for 2024

As I look ahead to 2024, here are some of my own predictions for the AI landscape, but they are only based on my personal knowledge and what I’ve seen so far:

This one’s a safe bet - continued growth and competition in the language model space
The rise of AI-generated art and its impact on the creative industry that can cause us all to rethink the actual meaning of the word Copyright
Increased focus on AI ethics and responsibility, which has already been the main talking point within the EU
The emergence of new AI applications and use cases that can lead into job search, employee screening and customer support, so it’s high time to get familiar with AI to strengthen your job potential

So, there you have it, folks! A wild recap of the most impactful AI events of 2023. As we move into 2024, one thing's for sure: the AI revolution is just getting started! Stay tuned for more exciting developments, and remember to keep it locked here for all the latest in AI and consumer tech!

2023: The Year AI Went Wild