This year, organizations invested more in Artificial Intelligence (AI) than ever before. With some help from AI, they ensure business continuity and support employees in the field and at home. We saw many new developments and speculations about how AI will improve our common future. So what can we expect next year? BrainCreators, the provider of digital inspectors for smart cities, infrastructure, and sustainability, gives five predictions for AI in 2022.
Multimodal AI
2022 will be the year for multimodal AI. Multimodal AI means the simultaneous use of images, text,
sound, and any other type of data in AI-based solutions. Using multimodal AI, descriptions of images
can be automatically generated, or videos can be automatically indexed and compared based on text
and sound. As a result, there will be more natural interaction and communication between humans
and machines. Multimodal AI can be used to improve language services, for example. A translation
service can extract clues from spoken words and objects from video and images. Additionally,
multimodal AI can be applied to improve predictions. Medical diagnoses will benefit from combining data from different sources, such as radiological examinations, pathology results, surgical information, and other patient data.
AI moves closer to the edge.
Next year we’ll see even more AI applications deployed on edge devices. Watches, doorbells, and
household items will get technological components, allowing intelligent services to improve our lives.
There’ll be increased demand for intelligent IoT and Edge solutions in smart cities, infrastructure,
and manufacturing sectors. As a result, organizations can significantly increase their margins, improve customer experience and optimize operational efficiency. Examples include the automatic inspection of spaces, surfaces, and objects, the anonymous tracking of visitors, and the observation of materials, grounds, and buildings.
Fake will be indistinguishable from real.
Deepfake technology will trick the human eye flawlessly. Generative adversarial networks are deep
learning algorithms, a form of artificial intelligence that mimics some human capabilities. Deepfakes
are generated (fake) videos that look real, and the technology is often seen as potentially dangerous.
After all, the application can be used in many ways, manipulating opinions, blackmailing people, or
damaging reputations. Yet, Deepfake technology can also offer benefits. For example, consider using
Deepfakes in film production by giving a deceased actor a role in a sequel. Deepfakes can also be
used to synthesize realistic data to help researchers develop new treatment methods for diseases,
eliminating their reliance on patient data.
Big players strengthen their monopoly with AI.
The really big players (FAANG) will further strengthen their monopoly on large data sets used to train
deep learning models. Hence, some AI models can only be trained behind closed doors. Although
these companies sometimes open-source their models, this will not happen with the training data.
This will create an even more significant gap between companies and free research.
AI for crime prevention
AI will be increasingly used to detect and combat cybercrime. At the same time, cybercriminals are
also increasingly using AI to develop smart malware that allows them to efficiently carry out large-
scale attacks that have a higher chance of success. In 2022, this cat and mouse game will continue to evolve, and the digital race between cybercriminals, security companies, and governments will
continue to develop.
“Due to the pandemic, process automation has accelerated as did digital transformation in the past
year, to keep the global economy running. AI has played a pivotal role in this, and I expect we will
benefit even more from this technology in the future,” says Maarten Stol, Chief Scientist at BrainCreators. “Data and AI have therefore become indispensable in this strongly digitizing world.
Nevertheless, we are still in a transition phase, which causes growing pains. Therefore, we must
remain critical and involve more diversity and disciplines in developing AI before it is implemented. In this way, we can use the technology responsibly and take advantage of the many benefits that AI
offers.”