📅 A recap of 2023’s most important tech advances - Part 1

Welcome to this edition of Loop!

With the new year ringing in, I thought it would be the perfect opportunity to look back on what happened in 2023. There’s a lot to go over, so I’ve split the recap into 2 parts.

Since technology rarely advances in complete secrecy, there are always clues as to what lies ahead of us. Plus, it’s good to refresh our memory given that there have been so many announcements in the last year.

‏‏‎ ‎ HIGHLIGHTS ‏‏‎ ‎

• How ChatGPT has changed both the tech industry and the public’s perception of AI

• Timeline of the biggest technology advances

• Key themes from the year

Let's jump in!

It has been a year like no other, with Generative AI emerging as the dominant theme for 2023.

The tech industry has been galvanized into action by OpenAI’s release of ChatGPT, as companies started to compete against each other in the GenAI race and release new products.

It’s worth noting just how much ChatGPT has captured the public’s imagination and changed perceptions of AI more generally.

Previously, AI was seen as an abstract technology that only the big technology companies were using - whether it was to sell ads, recommend products, or for facial recognition.

But with the release of ChatGPT, AI has become much more tangible for people.

It’s now seen as something that can answer almost any question they have, summarise entire documents into text that can be more easily understood, or can even create new images and video clips.

It’s now something that they can see working in front of them.

I still believe there is a lot of misunderstanding from the public about what AI actually is and how it’s being applied in practice, but the dial is starting to gradually shift as more people try out these tools.

2023 was an incredibly busy year for the tech industry and it often felt like there was a new product being launched every day.

To get a good overview of what has happened in the last year, let’s refresh our memory on some of the big announcements:

• Microsoft launches the Azure OpenAI service

• CNET finds errors in more than half of its AI-written stories

• BuzzFeed says it will use OpenAI tools to “enhance” its content

• Microsoft invests $10 billion in OpenAI

• New York City schools ban access to ChatGPT over fears of cheating and misinformation

• Google announce that MusicLM can generate music from text descriptions, but won’t release it over fears of copyright infringement

• OpenAI release their own GenAI text detector

• Google announce Bard as their ChatGPT competitor

• Bard makes a factual error in Google’s first demo

• OpenAI introduces ChatGPT Plus at $20 a month

• Revealed that Google had previously invested $300m in Anthropic AI for a 10% stake

• Microsoft release Bing Chat

• Snapchat release MyAI feature to their paid users

• Bing Chat’s internal rules are leaked online

• Google release Bard to the public

• Meta’s LLaMA model is leaked online

• Humane announce they are working with Microsoft and OpenAI on using AI in a future product

• Microsoft lays off its entire AI ethics and society team

• OpenAI announces GPT-4, which is also multi-modal

• Anthropic launches Claude

• Runway announces a new text-to-video model, called Gen-2

• Adobe launch their own AI image generator, called Firefly

• GenAI images of the Pope in a puffer jacket go viral

• Top AI researchers write an open letter and call for a pause on ‘giant AI experiments’

• Italian regulators order ChatGPT ban due to GDPR concerns

• Meta release their Segment Anything Model (SAM), which is capable of one-click segmentation for photos and video

• Stability AI announce their own language model, called StableLM

• Google combine Brain and DeepMind into one team

• Family of F1 legend Michael Schumacher plans legal action over fake AI interview

• US Republicans publish an AI-generated attack ad

• Geoffrey Hinton quits Google to speak about the risks of AI

• Writers Guild of America (WGA) vote to strike over the use of AI for script writing & low pay

• Anthropic increase Claude’s context window to 100k tokens

• Inflection release their Pi chatbot, which is designed to be more personal

• Google releases MusicLM, after it previously said it wouldn’t

• Eating disorder helpline replaces their staff with a chatbot, which then generates 'harmful' responses

• Apple finally unveil the Vision Pro as their AR/VR headset

• AWS invest $100 million to help companies use generative artificial intelligence

• EU formally approves the AI Act

• NY lawyers are fined after they used ChatGPT to create a legal brief, which included fake case citations

• OpenAI sued by radio host after ChatGPT made false accusations about him

• Gizmodo publish their first AI-generated story

• Meta release LLaMA 2, partner with Microsoft

• Anthropic announce Claude 2

• OpenAI shut down their GenAI text detector, due to low accuracy

• Google’s medical AI chatbot is being tested in hospitals

• Sergey Brin returns to help Google with their upcoming Gemini project

• Google releases details about Robotic Transformer 2 (RT-2), which can control robots using natural language

• Google use LLMs to command robot dogs to walk and run

• Meta announce Code Llama for programmers

• AI-generated art cannot be copyrighted, rules a US federal judge

• Meta releases details about their multilingual speech translation model, called SeamlessM4T

• Google announce a watermark for AI images that can’t be edited out

• OpenAI releases DALL-E 3

• Microsoft will assume legal responsibility if their Copilot customers are sued for copyright infringement

• Amazon will invest up to $4 billion into Anthropic AI

• Stability AI releases a new text-to-audio tool, called Stable Audio

• Gizmodo’s owner replaced its Spanish language journalists with AI

• US military planning to fund fleets of drones that are ‘small, smart, cheap’

• Microsoft’s $69 billion acquisition of Activision has finally been approved

• OpenAI create a new team to assess the catastrophic risks of AI

• Microsoft is losing an average of $20 for every user of GitHub Copilot, per month - due to the high costs of running LLMs

• Cruise suspend all their robotaxi operations, after California agencies revoke their permits

• Meta starts to roll out AI-powered tools for advertisers

• President Biden issues an executive order to set AI safety standards

• UN creates a new AI advisory body for international regulation

• Google invests $2 billion in Anthropic AI

• DeepMind tease that their next version of AlphaFold is a “significant improvement”

• UK Government host their own AI Safety Summit

• OpenAI host their first developer conference, announce GPT-4-turbo and custom GPTs

• Humane launches the AI Pin, which is a wearable that’s powered by OpenAI models

• Sam Altman is fired as OpenAI CEO, then hired by Microsoft, then rehired by OpenAI

• Meta disbands its Responsible AI team

• Microsoft have used GPT-4 Vision to analyse TV episodes, live sports, and games

• Anthropic release Claude 2.1, which now has 200k context window

• DeepMind creates an AI tool that has discovered 2.2 million new materials

• Stability AI announce work on their own AI video generator

• Microsoft explore how GPT-4 could be used for radiology

• General Motors slash spending at Cruise by hundreds of millions of dollars, recalls their entire robotaxi fleet, CEO resigns

• Google launches Gemini as their rival to GPT-4

• EU reaches agreement on the upcoming AI Act

• Meta and IBM form an AI Alliance to promote open-source development

• Microsoft reveals work on their small language model, called Phi-2

• DeepMind use LLMs to make new discoveries in mathematics

• Cruise cuts 24% of their workforce

• Apple’s researchers have successfully run LLMs on devices with limited memory

By looking back on the year, we can see a few themes:

• Google’s shifting priorities with GenAI is seen clearly with MusicLM. Originally, they were against releasing it to the public - but decided to do so a few months later

• Microsoft are funding research into smaller language models, as the cost of running larger models is too high - which is leading to losses with their GitHub Copilot service

• Humane hinted they were working with OpenAI on a new device, which later turned out to be the AI Pin

• AI text detectors were popular at the start of the year, but quickly became exposed as unreliable. OpenAI would later pull their detector from the market

• Context windows for AI models have rapidly increased. Anthropic’s Claude has seen jumps from 9k, to 100k, and 200k today

• Companies that fired most of their staff and replaced them with AI models, would later find that they made a pretty big mistake. Many fell victim to the excessive hype and promises of new GenAI tools, instead of validating what they can realistically be used for

• Google, OpenAI, Microsoft, and Anthropic have all committed to protecting their customers from lawsuits over copyright infringement

This was a bit of a change from the normal format, but it’s a good time to review the year since there are very few announcements over Christmas.

If you want to read part 2, you can check it out here.

In it, we look at:

• A summary of the year, for each of the top tech companies

• The biggest trends we saw in 2023

• And what to look out for in 2024

Have a good week!

Liam

Feedback

Share with Others

If you found something interesting, feel free to share this newsletter with your colleagues.

About the Author

Liam McCormick is a Senior Software Engineer and works within Kainos' Innovation team.

He identifies business value in emerging technologies, implements them, and then shares these insights with others.