Investment in Artificial Intelligence broke all records in 2021, doubling the figures reached in 2020. The increase was 108%, reaching, according to data from CB Insights, 66.8 billion dollars. The number of unicorns born in this sector also multiplied by four, for which there are still good prospects in 2022.
Although the figures for the first quarter already anticipate that the figures for 2021 will not be reached, the brake on investment is much less than in other sectors. Artificial Intelligence continues to capture the attention of investors worldwide. Now, what are the factors that should not be lost sight of in 2022? What trends will mark its evolution in the remainder of the year?
The war in Ukraine, or the election campaign in South Korea, have meant that, for example, deepfakes become even more relevant, and that attention be paid again to detecting possible forgeries that could affect the politics or economy of a region. He too metaverse has become especially relevant in 2022, and with it the need for new AI-powered content moderation tools to monitor toxic behavior on-line.
These are the seven trends to watch closely in the last quarter of 2022:
1. Synthetic data to ensure privacy
Companies around the world are experimenting with synthetic data sets to enable data sharing and collaboration, while complying with the GDPR and other privacy laws. These data allow the time needed to have anonymized and quality data to be reduced from months to days.
These data are fictitious, that is, they do not identify any specific person. But, otherwise, they imitate the real ones in everything. For this reason, they are a very valuable tool during the development of Artificial Intelligence models: synthetic data makes it possible to dispense with the use of information from real users during the testing of the models, the stage in which their effectiveness is tested.
two. The race for chips
New innovations are about to change the rules of the game in the chip market. From 3D chips to chips in memory-computing. The rapid commercial success of AI across industries is creating a demand for hardware that can support intensive computing and workflows, both in cloud data centers and on edge devices such as cameras.
A market to which more and more actors are joining and in which it will also be necessary to take into account start-ups such as Cerebras Systems, which already offers what they define as “the largest chip ever built”, with 2.6T transistors and 850,000 AI cores.
3. The protection of virtual worlds
Spaniards spend an average of 659.6 hours a year playing video games, almost two hours a day, and the figure is even higher in the United States. There, 76% of minors play regularly. Given this scenario, it is normal for concern to grow about inappropriate content that players may find, especially minors.
To control them, companies are using Artificial Intelligence to detect harmful behavior in games and other virtual spaces, such as the metaverse.
Meta spent $13 billion on content moderation between 2016 and 2021, while TikTok claimed in 2020 to have 10,000 people working on these tasks. The concern that already exists in social networks has now been transferred to video games and virtual worlds.
Four. The fight against ‘deepfakes’
In 2022, the deepfakes or counterfeits have not only improved, but have become accessible to anyone with a smartphone and a little time. The large number of publicly available videos and the ease of using voice recordings means that the most famous people are constantly dubbed in recordings that many take for real.
The researchers warn that it will become increasingly difficult to distinguish faces, objects and videos from the real thing. detection technology deepfakes he will have to compete with the one who creates them, and it will not be an easy fight.
5. A new way to program
Artificial Intelligence is making coding easier and more efficient. Algorithms are taking natural language commands and translating them into computer code, signaling a new era for the development of software.
In June of last year, GitHub (acquired by Microsoft in 2018) and OpenAI (in which Microsoft owns a $1 billion minority stake) partnered to create GitHub Copilot. It is an assistant to write code based on Artificial Intelligence.
It doesn’t just make little suggestions, it’s a generative tool that tries to predict what you want to do and suggests full features with multiple variants. You can even write code from comments that are definitions written in natural language of what you need.
6. multimodal AI
This type of Artificial Intelligence can understand concepts from multiple modalities, such as video, text and 2D images, to improve search and generate content. This is the well-known case of DALL-E, which generates images based on a text message.
Until now, an Artificial Intelligence with images was capable of generating images, and one trained with text could create a text, but not combine all the concepts so efficiently.
7. ‘End-to-end machine learning’
In the fragmented AI market, platforms that manage a project from start to finish are emerging as a compelling option in 2022.
These providers end to end they can be a more scalable option for companies that want to build AI systems quickly and efficiently. In the increasingly crowded space of Artificial Intelligence, big tech and unicorns will continue to compete for market share if they are able to offer an increasingly wide range of AI services.
*** Alicia Richart Ramon is general manager of Afiniti.
Follow the topics that interest you