23 AI predictions for the enterprise in 2023

It’s that time of year again, when artificial intelligence (AI) leaders, consultants and vendors look at enterprise trends and make their predictions. After a whirlwind 2022, it’s no easy task this time around.

You may not agree with every one of these — but in honor of 2023, these are 23 top AI and ML predictions experts think will be spot-on for the coming year:

1. AI will be at the core of connected ecosystems

“In 2023, we’re going to see more organizations start to move away from deploying siloed AI and ML applications that replicate human actions for highly specific purposes and begin building more connected ecosystems with AI at their core. This will enable organizations to take data from throughout the enterprise to strengthen machine learning models across applications, effectively creating learning systems that continually improve outcomes. For enterprises to be successful, they need to think about AI as a business multiplier, rather than simply an optimizer.” 

—  Vinod Bidarkoppa, CTO of Sam’s Club and SVP of Walmart 

2. Generative AI will transform enterprise applications

“The hype about generative AI becomes reality in 2023. That’s because the foundations for true generative AI are finally in place, with software that can transform large language models and recommender systems into production applications that go beyond images to intelligently answer questions, create content and even spark discoveries. This new creative era will fuel massive advances in personalized customer service, drive new business models and pave the way for breakthroughs in healthcare.”

— Manuvir Das, senior vice president, enterprise computing, Nvidia

3. AI will completely transform security, risk and fraud

“We’re seeing AI and powerful data capabilities redefine the security models and capabilities for companies. Security practitioners and the industry as a whole will have much better tools and much faster information at their disposal, and they should be able to isolate security risks with much greater precision. They’ll also be using more marketing-like techniques to understand anomalous behavior and bad actions. In due time, we may very well see parties using AI to infiltrate systems, attempt to take over software assets through ransomware and take advantage of the cryptocurrency markets.” 

– Ashok Srivastava, senior vice president and chief data officer, Intuit

4. Open source ML tools will gain greater market share

“Next year teams that focus on ML operations, management and governance will have to do more with less. Because of this, businesses will adopt more off-the-shelf solutions because they are less expensive to produce, require less research time and can be customized to fit most needs. MLOps teams will also need to consider open-source infrastructure instead of getting locked into long-term contracts with cloud providers. Open source delivers flexible customization, cost savings and efficiency. Especially with teams shrinking across tech, this is becoming a much more viable option.”

— Moses Guttman, CEO, ClearML

5. Deep learning opportunities will boost demand for GPUs

“The biggest source of improvement in AI has been the deployment of deep learning — and especially transformer models — in training systems, which are meant to mimic the action of a brain’s neurons and the tasks of humans. These breakthroughs require tremendous compute power to analyze vast structured and unstructured datasets. Unlike CPUs, graphics processing units (GPUs) can support the parallel processing that deep learning workloads require. That means in 2023, as more applications founded on deep learning technology emerge to do everything from translating menus to curing disease, demand for GPUs will continue to soar.” 

— Nick Elprin, CEO, Domino Data Lab

6. AI will create meaningful coaching experiences

“Modern AI technology is already being used to help managers, coaches and executives with real-time feedback to better interpret inflection, emotion and more, and provide recommendations on how to improve future interactions. The ability to interpret meaningful resonance as it happens is a level of coaching no human being can provide.” 

— Zayd Enam, CEO, Cresta

7. Geopolitical shifts will slow AI adoption

“As fear and protectionism create barriers to data movement and processing locations, AI adoption will slow down. Macroeconomic instability, including rising energy costs and a looming recession, will hobble the advancement of AI initiatives as companies struggle just to keep the lights on.”

— Rich Potter, CEO, Peak

8. The role of AI and ML engineers will become mainstream

“Since model deployment, scaling AI across the enterprise, reducing time to insight and reducing time to value will become the key success criteria, AI/ML engineers will become critical in meeting these criteria. Today a lot of AI projects fail because they are not built to scale or [to] integrate with business workflows.”

— Nicolas Sekkaki, GM of applications, Data and AI, Kyndryl

9. Multi, hybrid-cloud MLOps and interoperability will be key

“As the AI/ML market continues to flood with new solutions, as evident by the volume of startups and VC capital deployed in the space, enterprises have found themselves with a collection of niche, disparate tools at their disposal. In 2023, enterprises will be more conscious of selecting solutions that will be more interoperable with the rest of their ecosystem, including their on-premises footprint and across cloud providers (AWS, Azure, GCP). Additionally, enterprises will gravitate towards a handful of leading solutions as the disparate tools mature and come together in bundles as standalone solutions.”

— Anay Nawathe, principal consultant, ISG

10. Advanced ML will enable no-code AI 

“Advanced machine learning technologies will enable no-code developers to innovate and create applications never seen before. This evolution may pave the way for a new breed of development tools. In a likely scenario, application developers will ‘program the application’ by describing their intent, rather than describing the data and the logic as they’d do with low-code tools of today.”

— Esko Hannula, SVP of product management, Copado

11. With spending down, AI will shift to practical applications

“This past year was filled with incredibly impressive technological advancements, popularized by ChatGPT, DALL-E 2, Galactica and Facebook’s Make-A-Video. These massive models were made possible largely due to the availability of endless volumes of training data, and huge compute and infrastructure resources. Heading into 2023, funding for true blue-sky research will slow down as organizations become more conservative in spending to brace for the looming recession and will shift from investing in fundamental research to more practical applications. With more companies becoming increasingly frugal to mitigate this imminent threat, we can anticipate increased use of pre-trained models and more focus on applying the advancements from previous years to more concrete applications.”

—John Kane, head of signal processing and machine learning, Cogito

12. ChatGPT will change the contact center, but not the way you think

“Chatbots are the obvious application for ChatGPT, but they are probably not going to be the first ones. First, ChatGPT today can answer questions, but it cannot take actions. When a user contacts a brand, they sometimes just want answers, but often they want something done — process a return, or cancel an account, or transfer funds. Secondly, when used to answer questions, ChatGPT can answer based on knowledge [found] on the internet. But it doesn’t have access to knowledge which is not online. Finally, ChatGPT excels at generation of text, creating new content derived from existing online information. When a user contacts a brand, they don’t want creative output — they want immediate actions. All of these issues will get addressed, but it does mean that the first use case is probably not chatbots.” 

— Jonathan Rosenberg, CTO, Five9

13. AI will drive the future of customer experience

“Digital engagement has become the default rather than the fallback, and every interaction counts. While the emergence of automation initially resolved basic FAQs, it’s now providing more advanced capabilities: personalizing interactions based on customer intent, empowering people to take action and self-serve, and making predictions on their next best action.

“The only way for businesses to scale a VIP digital experience for everyone is with an AI-driven automation solution. This will become a C-level priority for brands in 2023, as they determine how to evolve from a primarily live agent-based interaction model to one that can be primarily serviced through automated interactions. AI will be necessary to scale operations and properly understand and respond to what customers are saying, so brands can learn what their customers want and plan accordingly.” 

— Jessica Popp, CTO of Ada

14. AI model marketplaces will emerge

“Coming soon are industry-specific AI model marketplaces that enable businesses to easily consume and integrate AI models in their business without having to create and manage the model lifecycle. Businesses will simply subscribe to an AI model store. Think of the Apple Music store or Spotify for AI models broken down by industry and data they process.”

— Bryan Harris, executive vice president and chief technology officer, SAS 

15. Explainability will create more trustworthy AI

“As individuals continue to worry about how businesses and employers will use AI and machine learning technology, it will become more important than ever for companies to provide transparency into how their AI is applied to worker and finance data. Explainable AI will increasingly help to advance enterprise AI adoption by establishing greater trust. More providers will start to disclose how their machine learning models lead to their outputs (e.g. recommendations) and predictions, and we’ll see this expand even further to the individual user level with explainability built right into the application being used.”

— Jim Stratton, CTO, Workday

16. 2023 will be a major year for federated learning

“Federated learning is a machine learning technique that can be used to train machine learning models at the location of data sources, by only communicating the trained models from individual data sources to reach a consensus for a global model. Therefore instead of using the traditional approach of collecting data from multiple sources to a centralized location for model training, this technique learns a collaborative model. Federated learning addresses some of the major issues that prevail in the current machine learning technique, such as data privacy, data security, data access rights and access to data from heterogeneous sources.”

— David Murray, chief business officer, Devron

17. NLP plus object recognition will take search to the next level

“While most people write scrapers today to get data off of websites, natural language processing (NLP) progress has been made where soon you can describe in natural language what you want to extract from a given web page and the machine pulls it for you. For example, you could say, “Search this travel site for all the flights from San Francisco to Boston and put all of them in a spreadsheet, along with price, airline, time and day of travel.” It’s a hard problem, but we could actually solve it in the next year.”

— Varun Ganapathi, CTO and co-founder, AKASA

18. Advances are coming in real-time speech translation

“With remote work, boundaries are becoming increasingly blurred. Today it’s common for people to work and converse with colleagues across borders, even if they don’t share a common language. Manual translation can become a hindrance that slows down productivity and innovation. We now have the technology to use communication tools such as Zoom that allows someone in Turkey, for example, to speak their native language but allows someone in the U.S. to hear what they’re saying in English. This real-time speech translation ultimately helps with efficiency and productivity while also giving businesses more of an opportunity to operate globally.”

— Manoj Chaudhary, CTO and SVP of engineering, Jitterbit

19. AI-enabled phishing will grow

“By now, everyone has seen AI-created deepfake videos. They are leveraged for a variety of purposes, ranging from reanimating a lost loved one, disseminating political propaganda or enhancing a marketing campaign. However, imagine receiving a phishing email with a deepfake video of your CEO instructing you to go to a malicious URL. Or an attacker constructing more believable, legitimate-seeming phishing emails by using AI to better mimic corporate communications. Modern AI capabilities could completely blur the lines between legitimate and malicious emails, websites, company communications and videos. Cybercrime AI-as-a-Service could be the next monetized tactic.”

— Heather Gantt-Evans, CISO, SailPoint

20. Companies will turn to a hybrid approach to NLP

“In the year ahead, we will see enterprises turn to a hybrid approach to natural language processing combining symbolic AI with ML, which has shown to produce explainable, scalable and more accurate results while leaving a smaller carbon footprint. Companies will expand automation to more complex processes, requiring accurate understanding of documents, and extending their data analytics activities to include data embedded in text and documents. Therefore, investments in AI-based natural language technologies will grow. These solutions will have to be accurate, efficient, environmentally sustainable, explainable and not subject to bias. This requires enterprises to abandon the single-technique approach such as just machine learning (ML) or deep learning (DL) for their intrinsic limitations.”

— Luca Scagliarini, chief product officer, Expert.ai

21. AI-generated music will see advancements

“Advancements in AI-generated music will be a particularly interesting development. Now [that] tools exist that generate visual art from text prompts, these same tools will be improved to do the same for music. There are already models available that use text prompts to generate music and realistic human voices. Once these models start performing well enough that the public takes notice, progress in the field of generative audio will accelerate even further. It’s not unreasonable to think, within the next few years, that AI-generated music videos could become reality, with AI-generated video, music and vocals.”

— Ulrik Stig Hansen, president, Encord

22. AI investments will move to fully-productized applications

“There will be less investment within Fortune 500 organizations allocated to internal ML and data science teams to build solutions from the ground up. It will be replaced with investments in fully productized applications or platform interfaces to deliver the desired data analytic and customer experience outcomes in focus. [That’s because] in the next five years, nearly every application will be powered by LLM-based neural network-powered data pipelines to help classify, enrich, interpret and serve.

“[But] productization of neural network technology is one of the hardest tasks in the computer science field right now. It is an incredibly fast-moving space that without dedicated focus and exposure to many different types of data and use cases, it will be hard for internal-solution ML teams to excel at leveraging these technologies.”

— Amr Awadallah, CEO, Vectara

23. AI will empower more efficient devops

“When it comes to devops, experts are confident that AI is not going to replace jobs; rather, it will empower developers and testers to work more efficiently. AI integration is augmenting people and empowering exploratory testers to find more bugs and issues upfront, streamlining the process from development to deployment. In 2023, we’ll see already-lean teams working more efficiently and with less risk as AI continues to be implemented throughout the development cycle.

“Specifically, AI-augmentation will help inform decision-making processes for devops teams by finding patterns and pointing out outliers, allowing applications to continuously ‘self-heal’ and freeing up time for teams to focus their brain power on the tasks that developers actually want to do and that are more strategically important to the organization.”

– Kevin Thompson, CEO, Tricentis