In the year that is ending, artificial intelligence has taken us by surprise on several occasions and in several fields, including health, politics and art. There have been surprising and varied developments, and all indications are that we will be stunned a few more times in 2023.

The 7 key AI milestones in 2022

1. DALL-E 2 revolutionises imaging

It’s hard to pick the most shocking news of all, but the release of DALL-E 2 would be a good candidate. In April, OpenAI unveiled the new capabilities of its application to generate unpublished images from a description: an astronaut riding a horse on the moon or a pair of teddy bears working in a lab, for example.The fever broke out in June when two engineers decided to independently replicate this technology and make it available to anyone, under the name Dall-e mini, now known as Craiyon. From then on, it became popular and fuelled a race for the most advanced text-to-image AI programme, with Midjourney and Stability AI entering the race.

2. Deep fakes and actors with ‘AI rights’.

While we’ve already been reveling in and creating controversy over deep fakes for a few years now, 2022 once again left us with some examples of how the AI technologies needed to create them continue to evolve and move ever closer to perfection. And we’re not just talking about impostured images: the voice comes in too.

In May we saw the most talked-about deep fake of the year, the music video in which Kendrick Lamar adopted the faces of other celebrities such as OJ Simpson, Kobe Bryant or Will Smith, with astonishing verisimilitude. The level of quality of this technology has already made some actors think about protecting their rights, given the possibility of being synthesised and reproduced without their consent. This must be how Bruce Willis saw it, who in September was the first actor to sell his rights to generate a digital twin with AI.

3. The strange case of Blake Lemoine

The public was also stunned to learn that Blake Lemoine, a Google engineer, had been fired for revealing the secrets of Lambda, an artificial intelligence in the research phase, capable of emulating a conversation. The somewhat bizarre case began when Lemoine held a series of chats with the programme to find out if the machine was self-aware.

The engineer concluded that Lambda’s alleged rights were being violated and contacted a lawyer to defend her. After publishing the transcripts of these conversations on his blog – surprisingly enough – Google decided to fire him in July for sharing confidential information.

4. How to calculate 200 million proteins

In the area of biomedical research, DeepMind and the European Bioinformatics Institute announced that they had used AI to predict the three-dimensional shape of more than 200 million proteins from virtually every organism on the planet.

This achievement, made possible by the AlphaFold application, has been a boon to researchers around the world, who will now be able to more easily address their research in the fight against disease and other challenges such as plastic pollution, explained DeepMind founder Demis Hassabis.

5. AI in the fight against crime

Although not as international as in the previous cases, 2022 has been an eventful year in the field of AI for public discussion in the Netherlands. Police in the Netherlands used a deep fake in May to encourage citizen participation in the case of an allegedly murdered teenager.

But it also emerged that the Netherlands Forensic Institute was using deep learning models to find death threats among large amounts of communications tapped by police. It also emerged that a controversial algorithm of the Amsterdam city council was ranking young people allegedly more likely to offend, stigmatising these minors and even targeting them by the police.

6. ChatGPT

Although GPT-3 had been published since 2020, OpenAI decides to take a new step in the development of this AI model by converting the model into ChatGPT, a virtual assistant trained with millions of texts. Adopting the chatbot format, the programme becomes an immediate success and in the first week of December it gains its first million users.

This conversational assistant is able to answer complex questions (with a not inconsiderable margin of error), pose fictions, recognise its mistakes or program small pieces of computer code in various computer languages.

7. La Coruña, home of the public AI agency

And in Spain, to close the year, we have also experienced an important milestone for AI. Ahead of the upcoming European directive that aims to regulate this field, the Spanish government decided to create the AESIA (Agencia Estatal para la Supervisión de la Inteligencia Artificial) and designated in December that A Coruña would be the headquarters of the new body.

Its functions will be to supervise and guide companies in the development of this technology, as well as to penalise infringements of the regulations that will come into force over the next few years.

And what remains to be done

It is true that there are still days to go before the chimes, as this article is published. So it is not out of the question that something else will happen, worthy of entering the IA summary of the year. But it is certain that we can already look forward to a very interesting 2023, full of new developments, solutions, hypes and controversies.

What is certain to happen in the new year and beyond is that professions and profiles with expertise in artificial intelligence will continue to be some of the most sought-after and sought-after in the labour market. At the Foundation we are committed to providing the best free training so that, if you are interested, you can train for free and improve your digital skills in all areas of digital transformation.