A brief review of the usage prospects for companies, but also for individuals, in the year that has just begun
by Eugenio Zuccarelli
2022 was another year of exponential growth for artificial intelligence (AI), for both business and consumers. From corporate uses to user creations, AI tools have reached unprecedented levels in just a few months. However, there is still a long way to go to further promote the adoption of AI. At the moment, data scientists are the experts capable of better understand the uses that can be made of AI: However, leaders need to become the driving force behind a push towards greater adoption. Only when managers lead the charge will we be able to truly adopt these technologies in everyday life and see their fruits in companies. In 2023, we will see a leap forward in AI and it will encompass seven main areas.
1. New regulatory trends
In 2023, authorities and governments will have to face the complexity and the power of artificial intelligence systems. Over the years, these have become increasingly powerful, now reaching capabilities far superior to ours. As a result, governments will seek to impose controls on AI tools. This will be necessary to combat its potentials negative uses and ensure that users do not end up being discriminated against or disadvantaged due to poorly designed algorithms.
Right now, various countries are adopting different regulations. Some of the guidelines on AI in European countries are already starting to take shape and differ significantly from the regulations in other countries.
In the near future, every country or region will have the own regulations, aligning them with our core principles of privacy, data ownership, and general culture. For example, while we can expect an emphasis on protecting users' privacy and property particularly in Europe and the United States, the Chinese government will certainly move in other directions. It is therefore doubtful that artificial intelligence can advance in the same way all over the world.
2. Ethical implications
As AI systems become more powerful, their implications also become increasingly complex. For example, algorithms can now make accurate predictions, from cancer detection to autonomous vehicle driving. These systems, however, are only as good as the data used to train them. Moving forward, a will be applied check increasing to the most sensitive characteristics of the models, such as age, gender, and ethnicity, to ensure that the algorithm does not discriminate. Since the data used for most models have inherent biases due to historical reasons, action must be taken to ensure that such biases are not introduced into AI systems when making decisions.
3. Interpretability
Over the past decade, the priority of researchers has been to create ever more powerful and accurate models. We have finally reached a point, though, where their precision is extremely high, allowing us to venture into some of the more complex activities, such as autonomous driving. However, this came at a cost. Models have become increasingly difficult to interpret, as seen with deep learning systems, a clear example of how most algorithms are now "black box" systems, meaning not even their creators can fully explain why these have made a specific decision.
Over the next year, the focus will shift towards creating more interpretable models. AI systems can now perform calculations and predictions that previously could only be dreamed of. But, to ensure their adoption, we must promote interpretability. To ensure that algorithms gain the trust of executives, CEOs, doctors or lawyers we need to enable them to understand how these systems operate, to find out if they make decisions through human-like processes.
4. Generative AI for video and content
Generative AI is a type of artificial intelligence that generates new content or similar data, by style or content, to a given input. In the context of video and content creation, generative AI can be used in a variety of ways. We can see the strong interest in generative AI tools as happened with ChatGPT (Open AI), Lensa (Prisma Labs) and Stable Diffusion (open source). Models that are gaining immense popularity, with so many users generating images from text or creating new avatars from existing images that ChatGPT was able to amass 1 million in just five days. It's no wonder Open AI, creator of ChatGPT, DALLE-2 and other AI tools, estimates its revenue will reach $200 million in 2023 and over $1 billion by the following year, as indicated by Reuters.
Potential applications include:
- Generation of video or audio content: Generative AI is used to create video or audio clips after analyzing a dataset of already existing video clips. The algorithm can then generate new video clips similar in style to the original dataset. We can imagine various applications, for example in apps like TikTok, Instagram Reels and others.
- Written content generation: Similarly, Generative AI is increasingly being used to help generate written content such as articles, stories, or social media posts.
- Generation of graphs or images: Many applications are emerging in the creation of new images, providing artists with a "new kind of brush".
5. Distributed AI
Distributed AI uses various devices such as smartphones, computers or servers to perform tasks more efficiently by combining the computing power, memory and resources of multiple devices. It can be implemented in different ways, depending on the specific needs and requirements of the system, such as, for example:
- Peer-to-peer (P2P) distribution: in this approach, multiple devices or computers work together as peers to perform a task. Each device or computer contributes its own processing power, memory and storage resources to the task and the results are shared between all of them.
- Client-server distribution: In this configuration, a central server distributes tasks to multiple client devices or computers that execute the tasks and then send the results to the server, which combines the results and handles any additional processing as needed.
- Hybrid distribution: This approach combines parts of P2P and client-server distribution. Multiple devices or computers work together as if to perform a task, but some of them may act as a server to coordinate the distribution of tasks or to handle other tasks that require more processing power or storage space.
Distributed AI systems are useful for large-scale tasks in applications such as data analytics, machine learning, and some real-time applications.
6. 5G
Artificial intelligence will also make great strides thanks to the more widespread adoption of 5G in the next 12-18 months. Thanks to the data processing speed and higher bandwidth, it will be possible to generate an even greater amount of data via smartphones and other smart devices. The Internet-of-Things (IoT) will become one of the main sources of data for AI, especially in the health sector.
The improved connectivity of 5G technology will also enable artificial intelligence systems to communicate more fluidly than before. This should lead to more effective collaboration between the different systems. It is also expected that 5G technology will increase the capabilities of edge computing, helping AI systems get closer to the data source, enabling processing close to the source and thus unlocking privacy and data sharing issues.
This will lead to even higher adoption rates for AI-powered applications within enterprises. In fact, we are already seeing 5G applications with AI especially in the healthcare, manufacturing and transport sectors.
7. AI in business
Artificial intelligence has proven to be a great tool within companies to improve productivity and reduce tedious tasks. Benefits such as improved decisions through big data, speed and efficiency of processes, and reduced labor costs and human error are also on the way. We can expect to see various improvements in many areas, including:
- Customer care: here AI is mainly used to develop chatbots capable of answering customer questions and solving the most frequent problems in real time. When implemented correctly, they helped improve the overall experience, freeing up human reps to handle more complex tasks. However, chatbots still need to overcome significant hurdles before they become tools capable of providing a satisfying user experience.
- Marketing: AI algorithms can also help analyze customer data in marketing. This allows businesses to deliver customized marketing campaigns, helping to boost sales and customer satisfaction.
- Supply chain management: AI is often used to optimize logistics and forecast supply and demand. This helps support companies in their efforts to manage inventory with a high degree of efficiency.
- Predictive maintenance: By successfully analyzing equipment data you can predict when the next maintenance is due, reducing downtime and improving operational efficiency.
- Fraud detection: Algorithms analyze data to identify any financial transactions that could be fraudulent.
- Human resources: AI tools are capable of automating even more “human” tasks like resume screening and interview scheduling, allowing HR professionals to focus on more strategic work. However, as is often the case, this is a sensitive area where issues of bias often arise which can lead to discrimination and still need to be addressed.
- SalesAI is ultimately used to analyze customer data and identify potential leads, helping sales teams prioritize and act more efficiently by connecting only with customers who are expected to be the most interested and tailoring the products to individual needs.
AS SHOWN FROM THIS QUICK REVIEW, artificial intelligence is being adopted on a large scale, both by companies and by individual users. In 2023, this progress will continue even more accelerated, with many detractors and supporters of the technology, but what is certain is that AI cannot be ignored.
Eugenio Zuccarelli is a New York-based Data Science Leader, where he leads a team of Data Scientists at CVS Health, the number one health care company in the world and a Fortunes 500. Eugenio was ranked Forbes 30 Under 30, was a TEDx Speaker and studied between MIT extension, Harvard is Imperial College.
Article published in Harvard Business Review Italy of March 2023.
