Global investments in AI software and solutions are predicted to reach $126 billionOpens a new window in 2025 from $51.2 billion in 2022. The sector is nearing a new level of maturity, with investments skyrocketing and some world’s top firms exploring AI research. Thus, It’s high time for IT companies, large or small, to accelerate their AI deployment strategies. However, developing an ideal AI strategy often lands leaders in a muddle. So, here are some key insights from technology experts on how businesses can go about adopting AI to scale operations.
The question of whether or not to use AI has long been moot. With companies scaling up, incorporating AI into their operations has become unavoidable. As massive changes cost a substantial amount of money, labor, and missed output, it’s critical to evaluate how technology may be used to best benefit the business. The same may be said about artificial intelligence. Even though it has gone from a hype cycle to a “mature technology†industry, organizations who wish to employ it are still uncertain how to do it effectively.
To best know how to develop a robust strategy to scale up business using AI, Toolbox has gathered some great insights from within the technology industry. These tips can help IT decision-makers utilize funds allocated for AI in the most optimum ways.
Key Takeaways for CEOs Considering AI for Scaling Business
AI has moved beyond the hype cycle
Mike Loukides, vice president of emerging tech content at O’Reilly Media, says, “one part of the hype cycle that the Gartner model doesn’t take into account is that the number of trends that elicit hype at once is relatively small.†Loukides thinks the hype has now moved on to cryptocurrency and the metaverse. Large language models still get some hype, but even that is fading.
See More: ​​Five Things to Consider Before Investing in Trending Technologies of 2022
Loukides quotes Timnit Gebru’s interview with IEEE Spectrum, where Gebru said that if you ask how to use AI to solve problems, you are asking the wrong question. First, ask what the problem is. Then, ask yourself about possible solutions and whether AI is one of them. To do that, you have to answer many questions: Do we have the data we need? What will be the cost of retraining the model when it goes stale? What are ways in which an AI solution can go wrong? These questions should be asked about any technological solution, not just AI.
While Loukides suggests developing a blueprint before implementing an AI strategy and doing detailed research, Kumar Saurabh, CEO and co-founder of LogicHub, believes that AI is becoming practical in many areas, improving data analysis and accuracy of critical services like cybersecurity. For example, AI is already helping analyze billions of data points, find anomalies, and use machine learning to find the proverbial needle in the haystack.
Kaj van de Loo, CTO of UserTesting, believes that AI has shifted from the hype cycle and should be openly adopted by organizations everywhere. AI adoption is critical for companies who want to develop sustainable and healthy management of customer relationships, enterprise resource planning, and internal communication, he adds.
“We see most enterprise organizations moving up the AI adoption maturity curve; sometimes without being aware of AI powered experiences.†For example, when leveraging AI properly in the UX space, some companies automate AI to identify usability issues or automatically generate better design variations. Loo thinks that AI can enable new types of user experiences with other technologies, so it creates virtuous feedback loops that optimize both user experience and machine learning performance.
Big tech and SMEs should leverage AI more than ever, but how?
Kjell Carlsson, head of data science strategy & evangelism at Domino Data Lab, concurs that every company needs not one but multiple AI strategies. “The different methods, data and use cases for different types of AI applications are so distinct that there is little benefit and many pitfalls to creating a single AI strategy. Companies that have separate strategies, e.g., for natural language understanding (NLU), computer vision, chatbot, and recommendation engines, to name but a few, will be more successful than ones that attempt a one size fits all approach, he adds.Â
Carlsson thinks that all big tech companies have adopted AI more than their non-tech counterparts, but they do not leverage AI to more than a fraction of its potential. All have large parts of their business that have yet to leverage AI.Â
The most AI-advanced parts of business operations are still constrained due to a lack of professional-grade tools, scalable infrastructure, immature processes, and a lack of skilled data scientists.
– Kjell Carlsson, head of data science strategy & evangelism at Domino Data Lab
“Executives and line-of-business SMEs that are AI literate and understand how to effectively implement AI solutions into their organizations are the rarest and most valuable resource across enterprises today,†says Carlsson.Â
One idea coming out of the data-centric AI movement, which thought leader Andrew Ng has been spearheading, is that AI has often undervalued the role of subject matter experts. Loukides suggests the way forward in AI is getting better data, then AI developers really need to be talking to the SMEs. “They are the ones who understand the data, where it comes from, what biases it has, how it can be tagged, and so on. A lot of the manual labor of tagging and data preparation can be automated. To do that, you really need to in-depth understand the data you’re working with.â€
Rajesh Venkatachalam, chief marketing officer, ElectrifAi, highlights the current level of AI adoption in big tech companies and how exactly SMEs can leverage the increased expenditure on AI. He thinks that AI adoption is still in its infancy at big tech companies. Even with the limited adoption, the growth has been exponential though. “SMEs can sidestep the increased expenditure by using a targeted approach with business-specific last-mile AI, leveraging prior models, and domain knowledge. Such an approach does not involve platform roll-out, integration, and talent cost,†he says.
Identify and overcome the AI related issues
We have AI in place, but where does the problem lie in its implementation? The problems with AI have nothing to do with AI, points out Loo. We already have enough technological breakthroughs to transform most industries and generate new ones. The problems lie in “platforms, processes, and people.â€Â
Loo believes enterprises today have not put in place platforms that enable data scientists to be productive — to access the data, tools and infrastructure they need and to be able to collaborate. They do not have mature processes to monitor and maintain the AI solutions, let alone develop and operationalize new ones. “Above all, they do not have the leaders and managers that can understand, evaluate, implement, and oversee these solutions.†Some enterprises are starting to overcome these challenges, and there are breakthrough use cases that are spreading rapidly, but we will be “in a target-rich environment†when it comes to AI for at least the next few decades.Â
Saurabh thinks that the tech industry will benefit from less hype around AI, more transparency about how AI/ML is being used in practice, and end-customers benefit. For example, there is a huge shortage of skilled security experts, and conventional practices rely too much on manual analysis and chase down false positives. He suggests that effective AI systems can dramatically improve the accuracy and efficacy of detection and automation of routine decision-making, which is critical to staying on top of threats.
See More: ScaleUp:AI Conference 2022: Four Key Takeaways From the Premier AI Summit
Loukides makes a point about how companies need to use AI. “Maturity has a lot to do with how you are using AI, not what it’s doing.†Before setting ideal practices for AI, companies should ask a few questions: Are you using AI to give you good results, minimize the possibility of harm, and allow people to be more creative? Have you put in place the tools that will tell you about the quality of your results? Are you testing how your AI systems perform on different population segments?Â
“That’s the kind of maturity we want to see. This thought has a lot to do with Gebru’s point about starting from the problem rather than looking for ways to use AI,†he avers.
Key points for companies planning to deploy AI
- Be wary of black box claims
Saurabh suggests that businesses should be wary of ‘black box’ AI claims and look for solutions and services that can explain exactly how AI and machine learning are being effectively applied.Â
- Stop thinking start doing
Loo thinks the issue is no longer whether companies should adopt AI but how they should do so. As AI usage becomes more standard for companies, those who don’t begin to prioritize AI adoption will see products gradually become irrelevant. There is no substitute for software that engages and learns from users. Human-initiated interaction models will continue to fall behind as enterprise software with AI-powered features become the industry standard.Â
- Break the myth
It’s a myth that your company is not ready for AI, believes Carlsson. The range of methods and applications is so broad that every company, no matter how unsophisticated, can take advantage of them. AI is also more important than ever. AI models enable you to detect, innovate and respond to the rapid changes in your business environment, and those changes will only happen more rapidly. Transforming your business with AI isn’t easy, but not adopting AI is impossible.
- Spot companies’ AI need
Loukides analyzes that companies thinking of adopting AI practices have a lot to think about. Questions their organizations should pose include: as you grow towards a mature AI practice, are you building the tools that make life better for people? Are you trying to use AI for its own sake, or are you trying to use it because it’s the most appropriate tool to solve your business problem?Â
Has your organization increased its AI budget in 2022? Tell us on LinkedInOpens a new window , TwitterOpens a new window , or FacebookOpens a new window . We’d love to hear from you!