Skip to Content

4 Trends in Scaling AI for the Coming Year

Person in a data center, checking info on their laptop. trends in ai, artificial intelligence trends
Make sure your organization has a competitive edge when it comes to artificial intelligence (AI).[jamesteohart/Adobe Stock]

Learn the four areas of focus that will give your company a competitive edge as you scale.

In a world where there are at least 175 billion nuances to speech, how we speak – not just what we say – becomes all the more important. When dealing with robotics and natural language processing, these nuances make the difference between success and entry into the uncanny valley.

This is one of many trends coming up in the new year. Others, related to artificial intelligence (AI), will profoundly alter the way we live, work, and interact with technology. 

Since Brad Porter, CTO at Scale AI and Corey Patton, co-founder and CEO of Pramana Labs are breaking new frontiers for what’s possible in AI, we ask them to weigh in on the trends affecting the industry. To listen to their full episodes and keep up with other emerging trends, subscribe to our free IT Visionaries podcast with host Albert Chou.

Here’s what we found:

1. MORE IS MORE: Accurate AI comes from abundant context

“Paris Hilton is the name of a person,” said Chou. But what about someone who wants to stay at a Hilton Hotel while vacationing in Paris, France? “You can actually train these tools to recognize these differences, to recognize one’s an organization,” he said. 

The answer is, you ideally do both. You try to get a lot of data and you try to label it incredibly well.

Brad Porter, CTO Scale AI

Porter agreed. “Deep learning models, at the end of the day, are statistically reasoning from the prior examples they have seen. You need to feed it examples. And so the challenge in machine learning is you have to find people who can do the work of labeling this Paris.” 

“So this is the classic trade-off,” he continued, “of how much do you invest in getting more data, versus how much do you invest in the quality of the data that you’re labeling and annotating? The answer is, you ideally do both. You try to get a lot of data and you try to label it incredibly well.”

There are, of course, one-off scenarios where AI cannot simply go with the trends. If a self-driving car sees a white blur, it may interpret it as a plastic bag. But your data needs to be smart enough to account for the one-in-ten times that plastic bag may actually be a seagull. “You do want to capture these rare events,” he said. “The more accurate the model, the more accurate the data – the more accurate the data, the more accurate the results.”

Pair low-code tools with AI

Reimagine how your teams work. Use point-and-click tools to automate any workflow, whether you’re creating a new emergency response or optimizing an existing process.

2. MIRROR HOW WE TALK: Anticipate conversation flow

“Humans learn by having multiple different conversations,” said Patton. They get a piece of information, they learn, they ask a follow-up question. Pramana’s AI helps synthesize results in a way that already anticipates the person’s needs. Rather than just giving a one-word or singular response, Pramana answers in multiple paragraphs, with charts and stats that give context in normal, everyday prose. “We provide both ends of the spectrum for users of the tool,” Patton said.

Exceeding expectations is becoming increasingly important for consumers today. Natural language processing (NLP) is the linchpin to make this happen. With estimates of up to 100 trillion data points to consider, understanding precisely what the user is saying and then being a step ahead for the response will make the difference between AI that simply answers versus AI that impresses.

“It’s a two-way conversation,” said Patton. “We allow the user to ask a question in free text, but we also have a full end-to-end pipeline … that scans your database and pushes out narrative prose. You need to work with the consumer; it all starts with them.”

3. AI AS AN ASSET TO SOCIETY: Make people’s jobs easier

“There are huge opportunities to really help people,” Porter said. “This problem of high-quality data isn’t just restricted to these kinds of future sci-fi drones, but also to day-to-day challenges like ecommerce.”

Higher and higher fidelity of information leads to improved services in the marketplace. Chou pointed back to the analogy of a Roomba vacuum – without it learning to distinguish between what is dust versus what is your cat, it might pick up the wrong things. “It’s still learning,” he said, as is every iteration tool in AI.

These improvements don’t just help the consumer; they help the employee as well. With 4.4 million people leaving their jobs in 2021, it’s important to keep the workforce of your company as content as possible. One way to do this is by incorporating AI into the everyday tasks of employees so they can be less bogged down by the mundane and more confident in their data results.

4. AI AND CYBERSECURITY: Stay ahead of threats

“You need to see what’s coming before it hits you,” Patton said. Cybersecurity threats are getting more and more advanced and AI needs to be incorporated for best protection against attacks. 

In a previous interview with security expert Marla Hay, we learned about ways to bake security into the very design of a system from its inception. We also talked with four different experts in the field about how to cooperate in automated protection programs across the globe. With threat actors becoming more and more advanced in the field of security, anticipating problems is crucial to maximizing AI’s potential. 

“The problem was,” said Chou on the topic of trending, “it was never big enough to the total conversation that would make [people] believe something was happening.” Artificial intelligence, on the other hand, can foreshadow where to cast the conversation on prevention. 

AI is on the move. “What problems do you want to solve?” asked Chou, to “keep pushing the pace of innovation in your space.” Within each of these four areas of focus, there are multiple story threads to follow with where artificial intelligence is headed. 

  • Data labeling
  • Natural language processing
  • Low code and the employee experience
  • Cybersecurity

Each of these has its niche of experimentation and refinement as we continue to grow. But an underlying thread that Patton summed up well – no matter the industry – is it’s essential to continue to stay curious: “Here is a pain point of pinpoint accuracy that you can then solve. And the solution really provided an initial benefit, then the benefits grew and multiplied.” 

Porter added: “You have to do great for your customers. You have to do great for your employees. And if you can solve for all that and make a viable business … you can become an incredibly successful company.”

Responsibly create artificial intelligence

Remove bias from your data and algorithms to create ethical AI systems at your company.

John Kucera SVP, Product Management

John Kucera leads the Automation Services product team, directly responsible for Einstein Chatbots, Flow, and Einstein Next Best Action. He is also responsible for driving the Einstein Automate vision across the Platform, Mulesoft, and Salesforce Industries, enabling end-to-end automation, integrated across any system.

More by John

Get the latest articles in your inbox.