Apart from new and profitable AI based products, a few AI topics that could correlate with financial gain/loss in near future:

1. GAN based fake synthetic data designed to tilt decision makers and/or the software used by decision makers in predetermined direction

2. Replacing low-level decision makers / resource managers (entry level non-technical managers) with interpretable AI based software

3. Direct and custom-fit application of AI techniques to increase efficiency and lower cost

4. Direct application of AI to human resources departments: fair and objective recruitment, employee performance evaluation

Classification, Clustering, Object Detection are important tasks in the field of AI.

The current important topic in AI is Modelling based learning via iterative or deep learning algorithms.

Apart from new and profitable AI based products, a few AI topics that could correlate with financial gain/loss in near future:

1. GAN based fake synthetic data designed to tilt decision makers and/or the software used by decision makers in predetermined direction

2. Replacing low-level decision makers / resource managers (entry level non-technical managers) with interpretable AI based software

3. Direct and custom-fit application of AI techniques to increase efficiency and lower cost

4. Direct application of AI to human resources departments: fair and objective recruitment, employee performance evaluation

I fully agree with you https://www.researchgate.net/profile/Toygar-Akguen As additional direction to your list I suggest use AI for the forecast of the market and predicting of market gamer behavior on the basis of Big Data analysis of the history they preferences.

Full disclosure, I work for a data analytics company called Splunk, but will try to not push my company's products in this answer. We are providing ML tools for our customers to look for understandings that are hidden in the low-level, unstructured data that is continuously generated by the wide range of digital systems spread across a large enterprise.

Standard statistical analytics and correlative searches across large streams of unstructured data have been very useful, but applying ML is driving new use cases. Leveraging ML techniques such as regression, classification, clustering, anomaly detection, etc. on streams of unstructured data, some of our users are: building prediction models that adjust in real-time, enhancing forecasts with related data flows, watching time-based ebbs and flows of clusters as various data flows change, etc.

In summary, building ML into operational workflows enables businesses to respond much more quickly and effectively to the dynamics of their business. They can also enhance and optimize workflows based on near-real-time predictions and forecasts generated by applying ML techniques to the data flowing through the infrastructure of the business.

This is on top of all the other answers.

I was fortunate to sit on the investment committee with the group CEO and CXOs. One of the business-critical problems of deploying AI in the strategic decision-making process is that almost every AI is a kind of black-box algorithms. Accountability and Interpretability is a key part, and business owners do not want to take the risk over critical strategies that drive the success of the company.

XAI (Explainable AI) research is addressing the problem and sooner will be seen in core areas of business.

You can also read about Transformers, GAN, Spiking neural networks, and quantum AI that are becoming hot talks.