AI in 2023 (and Beyond) – What Will the Future Bring?
Artificial intelligence has come a long way in just a short amount of time. For reference, in July 2021 a survey showed that 84% of Americans were illiterate about AI. In less than two years, however, AI has become a household name, owing largely to the popularity of image-generating software driven by natural language processing that we have all witnessed.
So, where do we go from here?
It's not easy to say. But we're definitely not slowing down.
Stability AI, the parent company of the image-generating software, Stable Diffusion, has reached unicorn status after it raised $101 million in a seed round. This has been a watershed moment for artificial intelligence industries, as this much money has never gone into developing AI at such an early stage before.
We will most likely aim even higher and achieve even greater things than before. The interesting part? The first markings of this future progress can be seen at this very moment.
Image-Generating Software – But 3D
On December 16th, DALL-E's creators from OpenAI released Point-E, a DALL-E-like project designed to be a tool in 3D modeling. While it is lagging behind its competition in terms of resolution, it is hundreds of times faster. Yes, you read that right, hundreds of times!
From CGI in movies to NASA's moon mapping missions, 3D modeling has seen plenty of uses. Once OpenAI refines Point-E and increases the quality of the models that it produces, whole industries might experience the revolution that DALL-E had brought into digital art. What's more, the entire project is open-source and available on Github.
The Mayflower Autonomous Ship
The Mayflower Autonomous Ship (MAS) is the first fully autonomous maritime vessel with scalable technology that is able to sail across the Atlantic. And it did so, without humans, "manned" only by an AI captain on June 5th 2022.
With 6 AI-powered cameras, over 30 sensors, and 15 Edge devices to guide it, MAS has turned out to be a breakthrough that might just as well reinvent international shipping as we know it. Not only has the AI proved to be capable of handling split-second decisions at sea, but it has done so with respect to maritime laws and in a transparent decision-making process. This is vital for both the purpose of research and fostering trust and receptiveness of the global market towards this innovation.
In the near future, we might see autonomous vessels that save time and costs, successfully tackle the ocean's most difficult problems, and reliably suggest solutions, but also help develop the decision process of AI in general. Perhaps even in 2023.
Brain Dynamics and Liquid AI
When MIT researchers showed that neural networks work even faster when they mimic the brains of smaller species, it proved to be groundbreaking for dynamic, safety-sensitive tasks such as flying and driving. This year, however, they expanded on their discoveries to solve a century-old differential equation problem and unlock "liquid" AI.
The "liquid" part here denotes neural networks that are recurrent, meaning that they process data in time series. However, as the size of such networks increases, the number of simulated synapses of this "brain" grows exponentially, which capped the potential size of these liquid neural networks. That is, until this year.
In 2022, a team of scientists solved a century-old differential equation that explains the interaction of two neurons traveling between synapses. As a result, AI no longer needs to perform calculations in a step-by-step process but forms a closed-form continuous-time model that derives solutions from any point in time within the model. Technical jargon aside, what does this boil down to? Much fewer nodes of simulated neurons, and much more speed, and performance.
This will come in handy, particularly with weather forecasting and healthcare devices, e.g. in pacemaker monitoring, but also quite literally everything else that is time-sensitive. In other words, expect even more powerful, sophisticated, and accurate AI in 2023, and a paved road towards a general improvement in AI technologies by orders of magnitude!
Who among us has never thought how awesome it would be if they were able to talk with their cuddly best friends? With AI, this fantasy is soon to become anything but.
Aza Raskin, the co-founder, and President of the Earth Species Project described the idea behind the project. The gist of it is that language can be portrayed using shapes, which is a form that can be easily analyzed by AI. These shapes may also be made using animal "languages", which can result in the AI being able to translate the meaning behind dog barks, cat meows, or whale whistles. Furthermore, the AI will be able to translate our language into animal language, allowing the translation process to go both ways. It won't be much different than using translation apps to communicate with people when abroad!
How soon? Scientists say in the next 12, 36, or 48 months. That little is all it takes to turn fables into reality.
Skippet enables you to build apps, organize data, automate workflows and manage your life using generative AI. From a simple description of your problem, such as “I want to manage my bills”, Skippet’s AI will help you create an app based on your input, tailored to your needs. Plus you can use AI to update your app, view portions of your data, and more.
Skippet will be available in beta from February 2023 so make sure to sign up for our waitlist!
What was mentioned above is just a fraction of projects that we know of and can expect some results from in the year that is to come. It's just the tip of the iceberg, and it's just this year. What will come our way in three years? Five? Ten?
The journal Science’s December 2022 issue had a piece about DeepMind's AlphaCode. The article documents how a program has, for the first time, been able to match the programming skills of an average coder in standardized contests.
Inventing proteins, automating intercontinental shipping, talking with animals… With AI, as we say goodbye to the old year and greet the new one, it seems as if not even the sky is the limit. The options are so vast that they can almost seem staggering.
But we're not going ahead of ourselves. Even while we progress forward, we double-check our direction and place regulations to make sure we do not swerve from the correct path. Ethics are at the forefront. We have formed a UNESCO project dedicated to ethics in AI. We're discussing what's right. We are working against the remnants of historical biases that still linger in our data. We are teaching AI about our morals.
So you do not need to be afraid of the future. Not in 2023. Not after 2023. If we commit our trust and efforts to AI, the future will be bright and rewarding for both of us.
And the new year of 2023 most definitely – a happy one.