5 key artificial intelligence (AI) trends to watch in 2022

1. Improved language modeling

Language modeling is an AI that allows machines to communicate and understand human language. It can take a human language and translate it into computer code that can then be used to build, run and manage software and apps. 

This type of language modeling and processing used to rely on datasets with a few thousand examples that it could learn from. The newer language models have access to larger amounts of data, so they can be trained using millions (or billions) of annotated examples. For this reason, the big tech companies have started to champion it.

Google has created a language model called BERT that uses bidirectional tech like word2vec or GloVe so it doesn’t rely on context to understand instructions. BERT teaches itself the relationships between sentences by understanding sentence structure and corpus, in a way similar to a human mind. For example, if you give BERT two sentences (A and B), it can understand if sentence B makes logistical sense in relation to sentence A. Here’s an example:

Other language models like OpenAI’s GPT-3 use billions of language parameters and variables to process language instantly. Developers can build products by giving the platform a text prompt (like a sentence or phrase) that it can turn into code. It can also be used to get an instant breakdown of your data and gain valuable insights into your business. 

For example, if you asked GPT-3 what the biggest hurdle was for your customers when they are using your online checkout, it may reply with something like: “The main hurdle for customers at the checkout is the slow loading times. They also don’t have the option for single-sign-on or to use payment methods other than debit/credit cards.”

These valuable insights can help you make tweaks to your website to give your customers a better experience.

2. AIOps is changing the way teams handle tech disruptions

Businesses now collect data on everything. We know what our customers buy, when our website is busiest and what products are selling the best. With all this data comes more complex technology and processes. Our IT systems are growing into monsters of their own. 

AIOps hopes to change all that. It helps tech teams improve their key processes and decision making and allows them to have better ways to analyze their datasets. Using a mix of machine learning (ML) and data science, AIOps helps IT teams find, troubleshoot and fix problems in their tech stack or if they’re experiencing lagging or other issues. 

How exactly does this help?

Think about the last time you had a system outage. It may have taken hours to fix, your team can’t work and your customers grow frustrated. An AIOps system can detect problems (or even patterns to predict that something may happen) and then use datasets and science to recommend how to fix them automatically. Instead of spending hours on a costly, manual fix to an IT outage, AIOps can easily help your tech team fix disruptions. 

AIOps also learns from every problem and outage and keeps insights on file. With every problem or change in the system, AIOps can predict shifts and spot potential issues in a tech stack as it’s tracking and monitoring everything in real-time – something that humans just can’t do.

3. Decision Intelligence is helping businesses make smarter decisions

Decision Intelligence (DI) is when AI is used in a commercial decision-making process, focused on outcomes in areas such as marketing personalization, demand forecasting and inventory optimization, to name just a few.

It’s different from “traditional” decision making as it removes human bias and can analyze huge datasets instantly to give smarter recommendations about what customers want. Companies are now using Decision Intelligence to improve how they:

• Sell products: By analyzing data, Decision Intelligence can help businesses see which sales and marketing activities are making the biggest impact and give accurate predictions for future revenue expectations
• Hire staff: HR departments can use Decision Intelligence to fast track the hiring process and evaluate candidates using apps and automated skills assessment tests
• Manage performance: Track individual sales performance and make targeted decisions on which products and trends your company should be focusing on to grow your revenue

It’s easy to see how Decision Intelligence can be a gamechanger for managing supply chains and making smarter decisions around products and pricing. When a leading UK multi-channel retailer wanted to introduce Decision Intelligence, it aimed to identify the optimal price to mark a product down to while helping to protect the product’s margin as they clear lines from the stock file.

Humans would take days to analyze datasets and make a decision, if at all. So, the brand turned to Peak’s Demand Intelligence solution to correlate its data and get a better view of product demand and price point. It combined data like the store’s website analytics, sales data and ERP data which gave the brand an accurate view of demand per SKU.

Thanks to this data, the retailer could use algorithms to produce a “perfect price range” for every product in their inventory without charging customers too much. This small change led to the brand discovering an opportunity to increase margin by $3m – approximately 1% of the retailer’s overall turnover. 

4. Emergence of no-code AI is making it accessible to everyone

One of the main barriers for companies that want to start using AI is a lack of skills.

As the AI for business market is still relatively nascent, it requires new skills to understand and successfully apply its algorithms and machine learning. One report found that a lack of skilled engineers was the top reason companies were yet to use AI, with 19% saying the skills gap was a “significant” barrier in last year.  The emergence of no-code (and low code) AI may solve this problem. 

The hint is in the name: no-code AI. Because it’s a code-free system, it usually relies on a custom-developed platform that companies integrate into their existing tech stacks so they can use AI instantly. Once the platform is integrated, it can start feeding off existing datasets and help with decision making, automating processes and creating marketing campaigns.

5. Multimodal AI is helping to model human perception

Multimodal AI is used to merge computer and conversational AI models to create a scenario that replicates human perception. That’s a mouthful.

Multimodal AI takes data like speech, text and images and mixes them with algorithms. Using the data, the algorithm creates a perception of a situation to act as a (very helpful and knowledgeable) assistant.  

Last year, Google released its Multimodal AI, multitask unified model (MUM), which uses datasets to train itself in different languages. Unlike other forms of AI, it doesn’t need to be trained to learn how to complete tasks and it’s multilingual – it can answer questions in 75 languages! 

Let’s say you’re going to the French Alps in December (lucky you!) and need help preparing for your trip. If you told MUM about the trip and asked for help, it would reply with everything from weather scenarios to fitness tips and gear recommendations.

Credit: Google

OpenAI has also released multimodel AI platforms like DALL-E that can generate images from text descriptions. It uses 12-billion parameters and datasets of text–image pairs to create an image based on its understanding of the text. It can even combine multiple items and bundle them together to create images. 

Want to create an emoji of a penguin wearing a blue hat…red gloves… green shirt…and yellow pants? No problem. 

This technology will make it easier for businesses to plan for situations and instantly create assets for marketing and sales campaigns without waiting for a designer.

Wrapping up

It’s always important for businesses to keep on top of technology trends to remain competitive – and AI is no different.

There are now multiple types of AI that are receiving billions of dollars in investment to improve the way businesses work and make decisions. AI now impacts everything from human decision making to automated processes and program development. 

Thanks to the emergence of trends like no-code tools and Decision Intelligence, AI is no longer just a tool for expert programmers and developers. Businesses like yours can take advantage of this tech and use it to make smarter decisions and grow your bottom line – and do it quickly.

Here at Peak, we help businesses leverage their data with Decision Intelligence to make great decisions, all the time. We help brands like Nike, KFC and ASOS enhance their systems and platforms with AI to get the most out of their ever-growing data –  read about some of our customer success stories here.