Five current trends in the AI industry every business leader should know about

And businesses are already using AI in a variety of ways. From AI reporting and intelligent automation, to the application of machine learning in business intelligence, new AI applications are emerging all the time.

So what AI trends are emerging to support these AI tools for business? Here are five trends in the AI industry that will impact how your organization can benefit from artificial intelligence.

1: Generative AI

Unless you’ve been stranded on a desert island for the last 18 months, you’re sure to have heard of generative AI – the AI trend that took the world by storm when ChatGPT was launched in November 2022.

As the senior editor for AI at MIT Technology Review, Will Douglas Heaven, put it, “Never has such radical new technology gone from experimental prototype to consumer product so fast and at such scale.”

Generative AI (Gen AI or GenAI) refers to deep learning models that can be used to create (or generate) new content including images, music, text, code, audio and video. Gen AI models are most commonly trained on large language models (LLMs) but, as we’ll see a little later, they can also be trained on other data types.

Some current and future use cases of generative AI include

• Written content augmentation
• Image and video generation
• Data discovery
• Content summarization and classification
• Customer service chatbot improvement
• Code generation and verification
• Synthetic data production

In a recent Gartner pollof 2,500 business execs, the most popular purpose of investing in generative AI was found to be improving customer experience and retention, with over a third (38%) choosing this option. Other purposes included revenue growth (26%), cost optimization (17%) and business continuity (7%).

Find out how European retailer Carrefour is using generative AI__and process mining to transform procurement.

2: Multimodal AI

Until recently, many AI systems were unimodal, meaning they used a single source or type of data to perform a specific task. Multimodal AI, on the other hand, uses a variety of data types and sources (including video, audio, speech, images, text and numerical data), and has the potential to produce more than one type of output.

This ability to use multiple types (or modes) of data makes multimodal AI more reflective of the way humans interact with their environment, using a variety of sensory inputs. It means AI has a more nuanced understanding of a situation and can draw more insightful inferences.

ChatGPT was originally a unimodal AI, using only text-based data from natural language processing (NLP). Now it has three multimodal features that allow image uploads as prompts, voice prompts, and AI-generated voice responses.

Use cases for multimodal AI

1. Robotics: Robots need to perceive and interact naturally with unpredictable real-world environments. By enabling them to use data from multiple sources, including cameras, microphones, GPS and a variety of sensors, multimodal AI will allow them to do just that.
2. Healthcare: Multimodal AI can improve both diagnostics and patient care by fusing a variety of data types. It can analyze and combine medical images from X-rays or MRIs, verbal and non-verbal cues from patient interactions, sensor data from wearable devices, and clinical notes, to provide precise diagnosis and more personalized healthcare.
3. Security: Multimodal AI can enhance surveillance capabilities by analyzing video feeds, audio data and sensor data to accurately detect threats. It can also process various data types to reconstruct and analyze incidents, resulting in better situational awareness and response.

3: Explainable AI

Explainable artificial intelligence, also known as XAI, is a set of processes and methods used to help people understand why an AI system reaches a particular decision, recommendation or prediction.

Machine learning algorithms are often thought of as ‘black boxes’ that are impossible to interpret. But, as businesses become more reliant on AI to make decisions, it’s vital they can understand and trust the results and output these algorithms create.

To unlock the full value of AI, businesses need a comprehensive explainability strategy. As explained in this McKinsey article, creating such as strategy is complex because the people that are using outputs from a single AI system may have very different explainability needs:

• A bank that uses an AI engine to support credit decisions will need to provide consumers who are denied a loan with a reason for that outcome.
• Loan officers and AI practitioners might need more granular information to help them understand the risk factors and weightings used in rendering the decision to ensure the model is tuned optimally.
• The risk function or diversity office may need to confirm that the data used in the AI engine is not biased against certain applicants.

Explainable AI has a variety of benefits. It allows developers to ensure the system is working as expected, while enabling the business to drive real value, build trust and adoption, and mitigate regulatory risks. It also enables those impacted by AI decision making to challenge outcomes.

Three elements of an explainable AI strategy

An XAI strategy usually combines three key elements:

1. Prediction accuracy, which can be checked by running simulations and comparing XAI output to results in the training data set.
2. Traceability, which can be accomplished by limiting the way decisions are made and setting a narrower scope.
3. Decision understanding, which involves educating people working with AI so they can understand how and why it makes decisions.

4: Ethical AI

Also known as responsible AI, ethical AI is a major concern across the globe. To illustrate this, the World Economic Forum launched the AI Governance Alliance in 2023, to provide guidance on the responsible design, development and deployment of AI systems. What’s more, the European Parliament is set to introduce the Artificial Intelligence Act, the world’s first AI legislation.

Until the AI Act and similar laws are passed to give more guidance on AI ethics, businesses will have to implement their own AI governance processes. Some ethical considerations for businesses to consider when embarking on AI initiatives include:

• AI bias: Does the AI’s training data contain any form of bias that could lead the AI model to unfairly discriminate against particular groups of people?
• Privacy: Does the application of AI compromise an individual’s privacy by making decisions that affect them, based on their personal data, without their knowledge or consent?
• Transparency: Is AI generated content clearly labeled as such, and can decisions made by machine learning algorithms be adequately explained?

5: Conversational AI

Conversational AI is a type of AI that uses machine learning and NLP to simulate human conversation. It powers AI assistants and chatbots, enabling them to recognize speech or text input, translate its nuanced meaning, and respond in a way that naturally imitates human interactions.

Conversational AI use cases

Expanded customer support: One of the most common uses of conversational AI is powering an AI chatbot that can perform a wider range of tasks than a traditional chatbot – and is therefore closer to a human customer support agent. An insurance company, for example, might have a conversational AI chatbot that can onboard new customers, check coverage, suggest alternative products, take payments, process claims, and generally answer policy-related questions.

Interactive FAQ sections: FAQ sections on business websites tend to offer generic answers that often fail to answer customers’ specific questions. They also become outdated very quickly and are time consuming to update. Replacing a traditional FAQ section with a conversational AI-powered assistant means customers can engage in real conversations and get their individual queries answered.