2023 was a turning point in the development of artificial intelligence, from machine learning to deep learning. Applications evolved from discriminative tasks, including text-based sentiment analysis and image-based facial recognition, to generative ones, like ChatGPT and DALL-E. More astonishingly, training data covered applications such as ready-to-present business reports and ready-to-execute software code, in addition to one-dimensional text and two-dimensional graphics. What technological advances are we likely to see in 2024, aside from advances in computer vision and natural language processing?
New connections between data and human creativity and cognition are being made possible by the application of deep learning. One instance is “deep dreaming,” in which imagined scenarios, frequently with fantastical elements, can be combined to create artistic visuals. For example, dragons can fly over winter landscapes according to the creator’s imagination, and they can combine biological qualities learnt from existing animal photographs. While computer-generated art is nothing new (just consider your favorite animated film), data-driven artifacts give the fantastical a realistic touch, and they can be created in a matter of seconds. All who can dream should be able to access art and entertainment.
When we consider human comprehension and understanding, we consider education and higher education. For a long time, psychologists have proposed that learners possess many forms of intelligences, such as visual, logical, etc. Efficient integration of these into personalized learning journeys has proven to be a difficulty. Deep Learning provides one example. It is easy to generate assignments in a variety of styles and difficulty levels. Analogously, grading can be automated. Everything is set up and ready. Individualizing not only assessments but entire learning journeys is the next step. It is possible to customize the topic to be emphasized based on the strengths and weaknesses of the receiving student. To every one, a personally chosen textbook.
Deep Learning also does exceptionally well at handling complexity. This is due to the technology’s capacity to process and synthesize many types of information, which is one of its distinguishing features. It is proving to be useful in sectors where multimodal data is available. Take finance, for example. Balance sheets, stock prices, analyst reports, fraud indicators, and many other formats are examples of data formats. It has always been useful to have predictive algorithms that combine all of this data into risk scenarios and market estimates. However, they may now be produced rapidly and utilize a large amount of data. We should be safer in spite of the financial and economic unrest.
The world’s two biggest democracies, the United States and India, will hold national elections in 2024. Deep Learning finds fertile ground in the complexity of electioneering. Political parties can analyze data from social media, for example, to determine which way the winds are blowing and then produce material to win over supporters and win over undecided voters. Psephologists can produce more comprehensive forecasts of election outcomes as well as more detailed projections at the granular level by examining historical voting patterns and demographic trends. But great power also comes with enormous responsibility when it comes to Deep Learning. It will also grow harder to distinguish between real news and phony news.
Everything in this points directly toward a new skill set based on machine learning. Developing user-friendly applications, frequently for mobile and edge devices, is one of the components. Other elements include: (i) the capacity to work securely with distributed and federated data; (ii) comfort levels with model workflows, especially on the cloud; (iii) building user-friendly applications; and (iv) sharing the approach and outcomes, or “explainable AI.”
Working with and developing huge language models, together with the tokens, embeddings, and transformers that go along with it, are already opportunities for the more seasoned; this will only get better.
Much has been written about how the new AI that Deep Learning has produced is already altering the nature of labor. It appears that impacted professions, including as music and literature, which were formerly seen as creative and non-technical, are affected. Similar to email and cell phones, professionals in all professions will probably need to have some experience with generative AI. It appears that a significant degree of basic upskilling is anticipated by 2024.
