3 AI predictions for 2023 and beyond, according to an AI expert

admin — Fri, 07 Apr 2023 07:32:32 +0000

The field of artificial intelligence (AI) has seen tremendous growth in recent years.
Companies looking to harness AI must overcome key societal concerns.
Key predictions outline how to derive value from responsible AI development.

If there’s one thing we know for sure as we look to the next year, it’s that institutions willing to embrace uncertainty – from market conditions to geopolitical turmoil and everything in between – will are the organizations best suited to serve their customers and employees. , and shareholders.

The field of artificial intelligence (AI) has seen incredible growth over the past 5 years as the field has provided new possibilities to reduce uncertainty by leveraging data to react respond quickly to the changing environment as soon as new data is available.

The technology and its benefits are no longer unknown to the majority, instead, many have witnessed first-hand the ability of AI to work quickly and efficiently in solving many high-level challenges. most pressing of society. We’ve seen it play a key role in delivering a COVID-19 vaccine at record speeds, helping hospitals identify and treat their “most at-risk patients,” and said More broadly, there is a significant reduction in the amount of human error in the data.

As we look to the year ahead, we think the ramifications of AI’s heightened social awareness, increased regulatory pressures, growing incentive to invest in the space, and how AI will continue to boost productivity staff may appear. Practical and applied AI concerns will become paramount to creating continued value from AI growth.

1. Raise awareness and ethical concerns

Algorithm trends have been a topic of growing discussion and debate in the use of AI. This is a difficult topic to navigate, both because of the potential complexity of identifying, analyzing and minimizing the presence of bias in the data mathematically and because of the social implications of determining What is “fairness” in the decision-making process?

Fairness depends on the situation, in addition to reflecting values, ethics and legal regulations. That said, there are clear ways to approach questions of AI fairness using data and models with guardrails, as well as suggested steps that organizations can take. implemented to mitigate undiscovered bias issues.

The biggest source of bias in an AI system is the data it is trained on. That data may have patterns of historical bias encoded in its results. Ultimately, machine learning derives knowledge from data, but that data comes from us – our decisions and systems.

Due to the ever-expanding use of technology and increased societal awareness of AI, you can expect to see organizations testing their systems and local governments working to ensure bias. of AI do not negatively impact their inhabitants. In New York City, for example, a new law will go into effect in 2023 to punish organizations that use AI-biased recruiting tools.

2. Legal pressure increases

Over the next year, I expect companies will face increasing regulatory pressure around their AI models. Regulatory changes could include requirements around both explanations for individual predictions as well as detailed records and historical and lineage tracking of how models were trained.

Increased regulation of AI will eventually be welcomed by the industry, as evidenced by 81% of tech leaders who say they would like to see increased government regulation in a recent DataRobot survey. However, the recent Blueprint on the AI Bill of Rights , which provides a set of five principles and related practices for protecting the rights of the American public in the age of AI, has prompted companies to act on the same issue. motion. More and more companies are becoming aware of the need to react with the possibility of converting voluntary guidelines into regulation in regulated industries and the potential costs of achieving reactive compliance over a period of time. short time.

Because of this, I predict most companies will need to invest in systems with model-driven governance. By investing in systems with the right railings, companies can continue to focus on technological innovation with the peace of mind that their systems comply with legal and regulatory obligations.

3. Invest more in space

In 2023, I expect investments in AI to continue to gain momentum, especially among businesses most directly impacted by economic and supply chain disruptions, as well as industry leaders. cities that are typically most likely to scale in AI adoption, such as financial services, retail, healthcare, and manufacturing. However, I also predict that, while some investments will progress, some AI technology trends will continue to be tested.

Looking at financial services, for example, I expect that use cases will shift to AI systems that can improve fraud detection accuracy and speed up costly reporting processes. effort. With growing expectations and the onslaught of security breaches, financial services need to secure a competitive edge with AI technologies that can help mitigate these adverse issues. Additionally, AI will help improve job satisfaction and free up employees to focus on adding value to customers.

Looking at technology trends, general AI is getting a lot of attention based on newly developed deep learning models (from OpenAI and others). However, I predict these models are still too new to be put into practice for most businesses because of a number of challenges. The first is the fact that it is difficult to guarantee their behavior on such essential issues as bias and fairness; despite efforts, current versions can easily be broken. This means that businesses will need to really trust the vendors of these models as they will have no hope of building or creating their own.

Adapting these models to the desired use cases is also difficult for most people. While I expect to see companies continue to work with generalized AI, I believe applications will continue to be tested for many businesses over the next year until business cases and return on investment are achieved. Their expected investment is better understood.

Overall, however, businesses are focused on building an organization-wide AI mentality by continuing to invest in the space and fully integrate AI into their operations (including the assessment of AI operations). new developments) would be better suited to handle market uncertainty and drive long-term success.

Artificial intelligence (AI) in image processing

admin — Tue, 01 Jun 2021 08:09:47 +0000

Efficient use of unstructured data from large amounts of image and voice data has always been a challenge for data mining professionals. Processing unstructured data often involves the use of deep learning algorithms, and these algorithms can be daunting for beginners. In addition, processing unstructured data often requires a powerful GPU and a large amount of computing resources. This article introduces an image recognition method using deep learning. This method can be applied to situations such as illegal image filtering and object detection.

This experiment generates an image recognition model using the TensorFlow deep learning framework in Alibaba’s Cloud Machine Learning Platform for AI. The whole procedure takes about 30 minutes to complete. After training the model, the system was able to recognize the bird in the following image and return the word “bird”:

This test can be generated from the following TensorFlow image classifier template:

Abstract

As the main means of human communication and understanding of the world, images are one of the important sources of information of human intellectual activity. With the development of the times, the demand for image processing technology is increasing day by day. The rapid development of computer technology also laid the foundation for the application of image processing. To achieve better image processing efficiency, this paper focuses on the application of artificial intelligence algorithms in image processing. Image segmentation is a technology that decomposes images into regions with different characteristics and extracts useful targets. This can be considered as a combinatorial optimization problem. It is possible to apply artificial intelligence algorithms to optimization problems. First, this paper introduces the ant colony algorithm in the artificial intelligence algorithm, builds the basic principle and mathematical model of the ant colony algorithm. Second, in order to improve the global search ability of the ant colony algorithm, this paper introduces the fish crowding function into the ant colony algorithm. Finally, the improved ant colony algorithm is used in image segmentation to improve the efficiency of image segmentation. Simulation results show that using ant colony algorithm in image segmentation is feasible. And the improvement of ant colony algorithm optimization is effective. The improved ant colony algorithm applied in image segmentation can greatly improve segmentation performance hiệu.

Lưu trữ Articles on AI and image processing - BEIT - Vietnam's leading technology, software development company

3 AI predictions for 2023 and beyond, according to an AI expert

1. Raise awareness and ethical concerns

2. Legal pressure increases

3. Invest more in space

Artificial intelligence (AI) in image processing

Abstract