As AI emerges from its early adopter molds and enters the mainstream, analysts foresee how the technology will take on more roles and change the landscape of various industries. Companies are already banking on AI assets for greater efficiency, faster insights, and enhanced customer experiences. Let’s look at what we can expect from the AI market as it matures in 2023.
Top 10 AI Trends to Watch Out For in 2023
1. Creative or generative AI
Generative AI refers to the sub-field of machine learning generating new data or content using an existing data set. Its goal is to produce something close to the original, real-world input data. This AI type uses deep learning algorithms to learn patterns and features in that data set, which may consist of code, text, images, audio, video, or other data types. Generative AI already has a wide range of applications. Here are three in-demand examples, all produced by the San Francisco-based AI research firm OpenAI, which will continue to shine in the coming year:
- Generative Pre-trained Transformer 3 (GPT-3)
Developed in 2020, GPT-3 is a language prediction model which “autocompletes” text after studying millions of web pages and scientific papers on the Internet. GPT-3 has 175 billion machine learning parameters. This generative AI product runs current copywriting tools, which generate human-like written content after you feed it with contexts, such as topics, descriptions, or introductory sentences. You can use this tool to develop outlines, summaries, essays, op-eds, and more.
However, GPT can contain bias because its output comes from previously published content, which can also have racial, religious, or gender bias.
ChatGPT is a bot version of GPT-3 that made its debut in November 2022. It’s a large language model which can answer questions and perform instructions after receiving “training” from human conversations and internet content written by humans. By studying human feedback sources such as Reddit, this AI “learned” what humans expect when others ask them a question—the “human style” of responding.
OpenAI designed ChatGPT to mimic conversational dialogues with humans. Because the bot can create and organize lists and human-sounding letters, industries foresee its wider use as an office assistant and customer service support.
Some of the concerns related to ChatGPT are its possible use for composing essays and academic papers. However, its generated content may include nonsensical sentences or even wrong information. Auditing ChatGPT’s performance may be necessary to prevent misinformation in customer care settings.
DALL-E became the most popular of the three OpenAI creations in 2022 due to its graphic-creation features. The product’s name comes from Spanish surrealist Salvador Dali and the robot in the 2008 Pixar animated film WALL-E.
You can create art by keying in a description, and DALL-E generates several versions. Or you can create a new image from an existing one by using text prompts. Users can do “in-painting” or delete parts of an image and replace it with something else. Or they can do “out-painting,” in which DALL-E can add more to an original photo (main subject or scenery). These capabilities make DALL-E a handy tool for the branding and creative marketing sectors.
OpenAI reportedly set policies that prevent DALL-E from creating “violent, adult, or hate images.” Nevertheless, this tool is also prone to bias, like GPT-3. DALL-E reportedly generated images of Caucasian men, following the prompt “the CEO.”
More DALL-E users might use the tool for creating animated art, specifically human-like images with a voice through AI-generated text-to-video platforms.
Other market leaders that have also developed AI tools include Amazon. Its text-to-speech tool, Polly, generates speaking voices for brands. The retail giant is also behind DeepComposer, which can expand a short melody into a complete song. Meanwhile, CodeAssist of Microsoft’s GitHub can assist developers in creating new software faster by completing codes.
2. Greater AI-human collaboration
The extent of AI’s support for various human functions has reached and will soar to new heights, earning them the name cobots or collaborative robots. Market insiders foresee more companies deploying machines with built-in AI to implement repeated and physically strenuous tasks. Doing so will allow human staff to perform more specialized duties. AI features can also enable teams to swiftly detect and respond to defects or failures,, improving safety and lowering costs for repairs or injury.
Cobots will be more widespread in these fields:
- Automotive manufacturing: car assembly, spray painting, surface polishing, systems checking, and the retrofitting or reconstructing car production lines to accommodate electric models. Companies with palletizing and welding activities expect to adopt more cobots with higher payloads and a longer reach.
- Agriculture: drones for seed planting, fertilizer and pesticide application, trespasser and invasive species tracking, and LED lighting and hydroponics for indoor farms
- Healthcare and hospitality: sample collection, hospital supplies restocking, surgery, injury recovery, health worker support in residential and nursing care homes for the elderly or disabled
- Food and beverage: warehousing, food packaging
- Electronics: quality inspection for phone chips, phone chip processors, and printed circuit boards
- Emerging technologies: torque sensors, proximity detection sensors, end-effectors (end-of-the-arm tooling such as vacuum, mechanical, pneumatic, and magnetic grippers
- Defense: clearing roads of explosive devices, sensors to detect explosives
Companies can also turn to these machines to ease labor shortages and issues in the supply chain. In particular, the healthcare, construction, and defense industries may replace traditional training methods with VR and AR-based learning for safety and reduced spending.
3. Ethics and regulation
Amid generative AI’s many benefits, people fear its misuse, such as the production of deep fake videos. Cybercriminals can use these tools to commit fraud, slander, blackmail, revenge, coercion, or extortion. Questions also arise regarding the boundaries of original and proprietary content. The AI sector expects users and customers to demand transparency, safety, and responsible practices.
The New York City Department of Consumer and Worker Protection has already passed an AI Law (New York City Local Law 144) that requires employers to meet bias audit requirements before using automated tools for evaluating job applicants. Moreover, hiring teams should inform candidates about their use of these tools for recruitment and job advertisements.
As early as 2021, the European Council has already submitted a proposal to regulate AI. The proposed legislation classifies AI applications and systems into prohibited, high-risk, and low-risk categories.
When approved, the AI Act will serve as the AI counterpart of the General Data Protection Regulation.
4. Democratization: low-code, no-code AI
The low-code, no-code trend in website and app development will carry over to AI, allowing organizations to customize these intelligent systems through pre-built templates and drag-and-drop methods. This way, AI’s integration into existing workflows will happen more quickly. AI usage will also scale faster within their corporate setup.
Besides using low-code, no-code AI for automating repeated tasks such as invoicing, form filling, and contact validation, businesses can program AI tools—like Sway AI and Akkio for data analysis of current processes and visualization of future performance.
AI market insiders also expect more cloud service providers to integrate AI into their offers due to its foreseen adoption in the long run.
Because IT modernization using low-code, no-code tools costs 70% cheaper and gets completed faster (as short as three days) than traditional methods, 66% of developers already use (39%) or plan to do so (27%) in 2023. Meanwhile, Gartner forecasted that by 2026, “citizen developers”— or those who did not take formal coding courses—will make up 80% of low-code tool development users.
5. Sophisticated cybersecurity
Another sad aspect of AI is that hackers can use it and its features to shorten the end-to-end lifecycle of their attacks from a few weeks to just days or hours, according to a McKinsey report.
As more industries adopt AI resources, critical infrastructure—including national civil infrastructure that supplies homes with power and water—may come under threat of hacking activities. At the same time, smaller, less protected organizations will continue to be vulnerable.
Career opportunities in information security will grow due to these new risks. Specialists can deploy and oversee security AI for:
- Data handling, including classification, cataloging, integration, and quality control
- Vulnerability management by surveying network traffic and identifying patterns suggesting criminal behavior
- Threat detection through predictive AI, which can project which of the thousands of alerts has the highest risks and deal with them first
IBM reported in 2022 that businesses with cyber risk management structures and policies saved an average of USD3 million and reduced breach lifecycles by 74 days due to quick detection and response.
Growing cyber threats may also push the insurance market to adopt new technologies and strategies to assess and manage cyber risks. Insurers may also introduce risk-based pricing and exemption clauses for ransomware and cyberattacks.
6. Digital twinning
A digital twin is a digital replica of an object or process in the physical world. Through AI, industries create virtual models for simulations, allowing them to predict how a product or system will perform.
The Omniverse platform of leading GPU manufacturer NVIDIA is an example of digital twin technology. Here’s how it’s helping the following companies:
Germany’s BMW Group uses the Omniverse as its virtual factory. The platform integrates data from various design and planning tools from different producers to create real-time, photo-realistic simulations in a single setting.
Staff from different sites and time zones can access this virtual space to plan or optimize details of production processes on demand, reducing the need for physical travel. The Omniverse simulates all of BMW’s 31 factories and all its elements—from its human associates and factory interiors to the assembly parts and robots.
American retailer Lowe’s Companies Inc. has also tapped the Omniverse to simulate its two stores: one in Washington (Mill Creek) and another in North Carolina (Charlotte). Personnel can access these simulated outlets using their desktop computers or the Magic Leap 2 augmented reality headsets.
The store’s Omniverse version will help restock shelves, reconfigure layouts, view product information from closed boxes on hard-to-reach shelves using “X-ray vision,” and optimize the customer experience through 3D heat maps indicating customer traffic and sales performance.
- HEAVY.AI (formerly OmniSci)
Omniverse enables the HeavyRF tool of analytics company HEAVY.AI to design the wireless network design plans of its telco clients. The AI tools simulate real-world environments, indicating the location of their customers and obstructions, including the material composition of the latter. This allows telcos to determine the best location for cellular towers and base stations for their 5G infrastructure, reducing site deployment costs and planning cycles.
Other examples include the creation of digital twin cities. For example, the Shanghai Urban Operations and Management Center has a digital clone of the Chinese city featuring its water bodies, airports, ports, and other establishments.
7. AI for personalization
eCommerce: While 62% of consumers expressed concern about AI bias, 69% of respondents to a Salesforce poll said that they’re open to its use by brands if it will improve their shopping experience. This trend will continue, given that 91% of consumers already interact with chatbots, the majority of which are AI-powered bots. AI marketing tools help brands analyze customer interactions to personalize product searches, recommendations, and messages.
Entertainment: AI reliance may also grow in filmmaking, where movie companies are already turning to audience analysis tools for finding the best stories for their next movies. Screenvision Media introduced its proprietary Cinelytics to its advertisers in 2022 while Warner Bros. adopted a similar technology earlier to predict ticket sales. As early as 2018, 20th Century Fox co-developed Merlin Video with Google’s Advanced Solutions Lab to forecast their moviegoers’ interest based on the AI’s study of movie trailers. Also, Netflix uses its subscribers’ watch history to suggest what to watch next.
Workplace: Worries about AI bias also exist among employers. However, market insiders say AI tools will continue to be an option for improving engagement through team communication software (such as Glint and Leena.AI) and workplace learning (using platforms like Hone and EdApp).
8. AI in voice technology
Voice biometrics: More businesses will explore using biometrics or voice authentication for identity protection in place of passwords or PINs. Voice assistants will make a “voiceprint” from your recorded sample and will use that to compare any new voice it receives whenever you unlock your device through speech. A growing number of banks are using voice recognition, allowing clients to access their accounts through speech authentication.
Voice cloning: AI can use a person’s voice sample to generate new audio. The technology will hasten the recording of voice-overs for a project and voice content for films, video games, and others. VoCapsule has a “voice bank” platform called My Legacy Voice. Members can access their voice data if they start having speech difficulties. Designated “primary recipients” can also access the data when the original member passes away.
Companies can also use voice cloning for localizing content, enabling people to hear promotions or instructions in their native language. Meanwhile, filmmakers can use this technology to manipulate an actor’s voice to speak different languages. They will transfer extracted elements from the artist’s original recording to a secondary track containing an interpreter or a voice talent’s speech. The process retains the accent and vocal performance of the secondary translation voice.
9. AI in motoring
The automotive industry foresees greater adoption of AI-based driver monitoring systems that can alert human drivers or activate autonomous driving if they detect drowsiness or illness. Adaptive cruise controls can send forward collision warnings and automatically adjust vehicle speed.
Also, manufacturers see automation—not electrification—as the future of driving. Renub Research forecasts that the autonomous vehicle market will skyrocket to $186.4 billion by 2030 from $4 billion in 2021.
Autonomous vehicle features continue to advance, from the presence of sensors and radars for object detection to convolutional neural networks. These networks recognize and classify terrain, paving the way for path planning, route optimization, and ultimately “training” self-driving cars to drive safely. Moreover, vehicle connectivity solutions are emerging that allow autonomous cars to “communicate” and avoid crashing into each other, pedestrians, and other objects.
10. AI in medicine
Precision medicine: As AI optimizes electronic health records, medical professionals can deliver targeted diagnostics, develop patient-specific drugs, and customize treatment plans. AI-enhanced diagnostics can reduce the harm one in four patients experience annually due to hospital negligence or oversight.
Virtual exams and decentralized clinical trials: Telehealth will stretch its powers to include remote physical exams with the help of smartphone solutions and wearables. Research and pharmaceutical entities can use the same devices in conducting clinical trials so that participants don’t need to travel to a trial site to answer surveys and assessments.
Emotional AI technology: AI with emotion recognition and generation capabilities will engage autistic children, depressed patients, and others with degenerative ailments like dementia.