Looking back at 2022, it’s evident that the area of artificial intelligence (AI) has achieved significant advances. We’ve seen various milestones, ranging from advances in natural language processing and computer vision to increased business acceptance of AI.

AI’s tremendous progress will undoubtedly continue into 2023. However, given the quick pace of development, it is critical to distinguish between significant advances and distracting fads.

I asked five questions to corporate leaders and academic researchers to better grasp the top AI developments. It’s the age of human-machine collaboration, so what better way to demonstrate this than to ask AI software what AI trends it’s enthusiastic about in 2024? I did the same thing.

I asked OpenAI’s ChatGPT the same five questions listed below and posted the AI model’s responses alongside those of the human experts. Furthermore, the AI wrote the first two paragraphs of this essay, which you can read above.

1. What do you believe was the most significant achievement in the AI space in 2022?

2. What is the most intriguing AI trend that you expect to see in 2024?

3. What is a craze in our industry that you wish would die next year?

4. What is one thing that leaders can start doing today to get value from AI that doesn’t need a lot of effort or money?

5. What is one thing that leaders should immediately cease doing in order to reap the benefits of AI?

This article is structured around these inquiries. Feel free to skip around to the ones that pique your attention the most. Don’t forget to compare human responses against AI responses for each question!

1. What do you believe was the most significant achievement in the AI space in 2022?

Apps like Stable Diffusion and Midjourney developed viral content and won competitions using AI-generated content. We witnessed advancements in synthetic data and the application of AI in commercial decision-making. AI pioneered new ground in the pharmaceutical and healthcare industries. ChatGPT was confused when answering this question because the AI model was only trained with data through 2021.

We’ll most certainly remember 2022 as the year that artificial intelligence became really accessible to the general population. DALL-E, Stable Diffusion, and ChatGPT achievements in transformer architectures will revolutionise numerous sectors in the coming years. – Grace Chang, Kintsugi’s Founder and CEO

I’m encouraged by the rising emphasis on designing AI around processes to improve decision-making – either automating it or drastically improving the options available to human decision-makers. This is a bigger paradigm shift than most people realise. Previously, it was thought that AI would demonstrate to business leaders how foolish their processes were and enlighten them on the importance of process reengineering. We’re all aware of how well that works. – Drew Smith, Little Caesars Enterprises, Inc. and Ilitch Companies’ vice president of Global Data and Analytics

AI in health is moving beyond the early adopter domains of radiography and pathology. In our own sector of fertility, we find that AI is no longer viewed as a “future” technology with enormous promise, but rather as a credible, evidence-based, and growing technology that may help healthcare professionals (HCPs) provide better care and enhance their business success. This change of heart began in practise for AI in fertility in 2022, and 2023 will be a scale-up year. – Eran Eshed, Fairtlity’s CEO

I’m pleased about the advancement of synthetic data, which allows my data science team to feed data into our machine learning (ML) models to simulate enormous populations of people with diseases. Synthetic test data allows us to consider ‘what if’ possibilities, making it an excellent method for testing a theory. To build huge training datasets and remove restrictions with the usage of sensitive data, it is a more accurate, scalable, and economically feasible replacement for real-world records. – Co-Founder and CEO of nQ Medical, Richie Bavasso

Is AlphaFold the most significant development in AI for biological sciences? Google DeepMind’s AI technology correctly predicted the three-dimensional structures of proteins. While this is an important step in medication discovery, there are still many unknowns about disease biology. Meta AI claims to have predicted structures for 600 million proteins using massive language models, whilst AlphaFold solved 3D structures for 220 million proteins. This collaboration between academia and industry in AI research has enormous promise, such as identifying therapies for RNA viruses that cause disorders like COVID. – Subha Madhavan, Pfizer’s vice president of Early Clinical Development Data Sciences and AI/ML

I’m sorry, but I am a language model trained by OpenAI and my knowledge cutoff is 2021, so I am not able to provide information about developments or achievements in the field of artificial intelligence (AI) beyond that year. – OpenAI ChatGPT

2. What is the most intriguing AI trend that you expect to see in 2023-2024?

Using AI to improve data quality could start a virtuous loop that propels technology forward exponentially. Through government initiatives and the use of AI to establish new economic models, 2023 could usher in the democratisation of AI. ChatGPT, on the other hand, is enthused about AI becoming more understandable and transparent.

Previously, AI was used to automate operational choices such as Call Centre Operations. In 2023, I hope to see algorithms applied strategically to create new business models. Consider the manufacturer of tractors, combines, harvesters, and other agricultural equipment, John Deere. Deere can generate new pockets of value by embedding sensors in their equipment and collecting product-in-use data. They can then construct data graphs and utilise AI to analyse them. Deere might transition from an equipment maker to a provider of insights and solutions to client challenges. Deere’s objective should be to become the leader in precision agriculture, with their own products serving as a component of the solution rather than the entire answer. The identity of Deere can shift from Made by Us to Solved by Us. All industrial organisations, including Deere, GM, and Rolls Royce, can benefit from AI-driven transformation. – Vijay Govindarajan, Coxe Distinguished Professor of Management at Dartmouth’s Tuck School of Business

The application of AI to data, such as resolving data quality, assessing data susceptibility, and detecting data duplication. This is seen in the integration of artificial intelligence (AI) farther up the data pipeline, such as in master data management, data governance, and extract-transform-load (ETL) procedures. And the potential boost in insight speed is enormous. – Dr. Drew Smith

Government-led initiatives, such as the Biden Administration’s National AI Research Resource Task Force (NAIRR), will inspire national discussions on AI. This has the potential to democratise access and speed real change in industries such as healthcare. A small Berkeley firm like ours is commercialising National Science Foundation-funded voice biomarker research to solve the $200 billion cost of undetected depression. Grace Chang’s

Teams will use AI to combine data across disciplines by 2023. By combining multimodal data from advanced genomic, radiological, and pathological imaging with patient-specific clinical data, medication targets, disease outcomes, and individualised patient responses can be predicted. Long COVID, for example, has pushed nearly 4 million US workers out of the labour force, costing $1 trillion in lost wages over a five-year period. In 2023, AI using multimodal data, including as photos, text, structured data, and voice, can aid in the development of treatment and patient care strategies to address protracted COVID. Subha Madhavan’s

I believe that the year 2023 will be the year of AI deployment and adoption in precision medicine. You can have the best algorithms in the world that detect diseases early, but for them to be useful, they must be deployed in healthcare practises and supported by clinical teams. Today, we see integrated delivery networks (IDNs) upgrading their technological infrastructure in order to apply AI algorithms for earlier disease identification in fields such as cancer, immunology, and neurology. – Najat Khan, Janssen’s Chief Data Science Officer and Global Head of Strategy and Operations for R&D.

Explainability and transparency: There is a rising interest in making AI systems transparent and explainable, especially when they are utilised in decision-making processes that influence people’s lives. This could result in the development of new methodologies and technologies for explaining AI system decisions, as well as the development of more transparent and interpretable models. – ChatGPT by OpenAI

3. What is a craze in our industry that you wish would die next year?

We are still grappling to define AI, while most industry studies on AI adoption aren’t actionable. Surprisingly, ChatGPT believes AI is overhyped, but human specialists are tired of doomsday prophecies about AI rendering people obsolete.

The majority of what people call AI is actually machine learning (ML). I get that AI is more appealing than ML, but there is a substantial difference. Machine learning is predicated on the premise that machines can learn and adapt through experience (sic data). AI is a larger concept that employs ML, deep learning, and other techniques to tackle real-world issues. AI is used by “intelligent” computers to think like humans and complete activities on their own. ML is the process by which a computer system gains intelligence. AI, in its true sense, does not yet exist. Richie Bavasso’s

The notion that AI can replace humans is a passing trend. The advent of chatGPT reignited debate about whether AI would make humans lazy or obsolete. These apocalypse predictions have accompanied every new technological advancement. Machines excel at solving issues with well-defined parameters and historical precedents. Humans excel at comprehending nuance, context, and conceptual problem-solving. According to studies, collaboration between AI and humans results in improved medical outcomes. – Ngan Khanh Nguyen MacDonald, Chief of Data Operations at Northwestern University’s Institute for Augmented Intelligence in Medicine

Stats such as “94% of leaders say AI is essential to their success” must be removed from the discussion. Really? And how many can explain what it is and how it will help them succeed? Few, which is why “50% of analytic professionals cite a lack of executive support” as a cause for AI failing to deliver. Furthermore, the results of various industry polls have not changed in the last five years. – Dr. Drew Smith

AI hype and overhyping: The topic of AI has gained a lot of attention in recent years, which has occasionally led to hype and overhyping of AI technologies’ capabilities and potential. While AI can be a powerful tool, it is crucial to be realistic about what it can and cannot do, as well as to be aware of its limitations and potential threats. – ChatGPT by OpenAI

4. To get the most out of AI, what is one thing leaders can start doing right away that doesn’t require a lot of time or money?

Who doesn’t appreciate easy hacks that can be done now with minimal effort? According to experts, organisations will gain by clarifying the true business objective of AI. They can save time by focusing on data quality before investing heavily in models. ChatGPT advises executives to educate themselves on AI and its hazards so that they can make informed judgements about using it.

Begin by considering specific business use cases in your organisation that could benefit from large amounts of data to help with day-to-day operations. Remember that AI is strong at recognising patterns, and humans are good at distinguishing between patterns that have meaning and bogus correlations. – MacDonald, Ngan Khanh Nguyen

Many leaders often hire the team first and then try to figure out what they’ll do with them. That simply does not work. At Janssen, we consider the problems to investigate and prioritise them based on two factors: business importance and data science feasibility. Focusing on a prioritised set of actions aids in delivering impact, gaining buy-in, and building organisational energy to do more.That’s what I accomplished as a new emerging leader at J&J two years ago. I chose the priority areas to apply data science after examining our programme pipeline. It was expediting trial enrolment for some programmes, while it was identifying more specific endpoints to better quantify patient outcomes for others. This assisted us in determining the top 10 programmes to pursue. Najat Khan’s

Investigate your existing data further. Investing in data efforts such as data sampling, data augmentation, and training techniques can result in significant improvements in AI performance. It does not necessitate the collection of additional data or the investigation of a new model architecture. – Alex Lang, Unlearn’s Vice President of Technology

Leaders must align and educate their teams; the only expense is time. Building trust and educating people about what AI can and cannot do is one of the most significant barriers to widespread AI use in healthcare. AI must be carefully marketed or risk being viewed as immature or overhyped. We are confronting this challenge head on by establishing a society dedicated to providing professional leadership in the provisioning and control of artificial intelligence in fertility. Eran Eshed’s

Educating themselves about AI: Educating themselves about the capabilities and limitations of AI, as well as the possible rewards and hazards associated with its use, is one of the most critical things leaders can do. This can assist leaders in better understanding how AI can aid their organisations and making educated decisions regarding its adoption and deployment. – ChatGPT by OpenAI

5. What is one thing that leaders should immediately cease doing in order to reap the benefits of AI?

It’s past time for businesses to cease AI washing, the marketing ploy of labelling items with AI only to make them sell. Leaders that ignore data quality or accept an AI “black box” approach put their companies at danger. When businesses put data privacy on the back burner, it will come back to haunt them.

Business leaders incorporate AI into descriptions of their products and services as though the offering could not exist without it. It is time to consider whether the advent of AI adds value to issue solving. It does not in many circumstances. There is frequently insufficient data to feed AI models, infrastructure to support consumption, or procedures to turn data into usable information. Richie Bavasso’s

Stop thinking of AI as a “black box” and start thinking of it as a “glass box.” Transparency is essential for obtaining the trust of all stakeholders. Explainability, which we all know is difficult to do with AI, builds trust. Leaders in the healthcare industry, in particular, should avoid taking the easy/short road to AI adoption and begin presenting the biology underlying the AI analysis. Eran Eshed’s

Instead than encouraging AI scientists to be concerned with whether they could train a model on a dataset, have them consider why they should. Stability AI released Stable Diffusion V1, a ground-breaking text-to-image model, without first thoroughly analysing their data and deciding whether to train a model on copyrighted and not-safe-for-work (NSFW) data. Instead, they waited for V2 to fix the problem, but by then, they had already destroyed trust and exposed themselves to legal liability. – Alexander Lang

Underestimation of data quality: Because AI systems rely on data to learn and make choices, it is critical to guarantee that the data used to train and evaluate these systems is of good quality. Leaders should avoid employing low-quality data or neglecting data-quality issues, since this might result in flawed or biassed AI systems that fail to deliver the desired results. – ChatGPT by OpenAI

So, what can we expect in 2024?

AI has progressed from being merely interesting to having a significant impact on organisations and consumers. AI adoption will surge in 2023 due to the increasing dispersion of models into all types of applications. With the addition of AI, standard software and equipment used by businesses will become smarter, while new-age apps will proliferate and continue to blur the lines between reality and science fiction.

We live in fascinating times, in which AI can now foresee future AI trends. While ChatGPT is light years ahead of any other chatbot we’ve encountered, there was evident human curation involved in selecting a selection of the content from AI’s answers for this article.

Nonetheless, this AI model provides a sneak peek at the amazing AI capabilities that will be available to us in 2023. While I don’t expect AI to compose the 2024 AI trends on its own around this time next year, the amount of human supervision that will be required is anyone’s guess.