7 ways to make the business at large comfortable with artificial intelligence

As those charged with the design, construction and deployment of artificial intelligence, from data engineers to developers recognize, AI is a powerful mechanism for amplifying human knowledge, skills and efficiency. But how can AI supporters use AI to correct a moribund or toxic corporate culture? This is probably the thorniest challenge with AI deployments.


Photo: Joe McKendrick

Entrepreneurs and experts on the front lines of the AI ​​revolution recognize that this is a barrier that technology alone cannot solve. Stephan Baldwin, Founder of the Assisted Living Center. “These are encouraged by principles that shape the day-to-day internal and external functioning of a business. ”

One of the challenges, Baldwin points out, we “artificial intelligence models operate on the basis of historical data, which means they are subject to the biases that we humans have when collecting information. Sometimes an automated process doesn’t take into account the people it governs. . “

So the challenge is to put people first in all AI projects. AI practitioners make the following recommendations to create a culture that is people-centered, but AI-driven:

Extend AI ownership and accountability beyond IT. AI needs to be a company-wide initiative, with all parties involved. “The successful and productive deployment of AI is a cross-functional effort far beyond simple data science,” says Dr. Michael Wu, chief AI strategist at PROS. “Extended teams need to go from the technical side, involving IT and cloud operations for data security and governance, to the business side, involving change management, education training, adoption, best practices. “

Recognize that AI is just code. It is not some mysterious black art capable of outwitting humans. “AI is no longer magic, and companies now seem to understand it,” says Beerud Sheth, co-founder and CEO of Gupshup. “AI isn’t trying to replace humans, but enables a more human conversation that has the power of automation and intelligence that a machine might have.”

Target AI in areas where it has the most impact. The best people in the business to promote and launch AI vary widely from industry to industry, Wu points out. “But the common theme is that organizations need to have a trusted source of clean, as a by-product of normal business operations, ”he says. and resolutions. Transaction data in sales organizations tends to be quite clean as it is necessary for good accounting practices. This data will continue to fuel their AI / ML as they learn. On the other hand, although marketing organizations have a lot of data too, they are often noisier and often require cleaning before they can be used in AI and ML production. ”

Sheth sees the most activity in customer support, product discovery, and employee engagement at customer organizations. “Considerable advancements in linguistic analysis and machine learning have enabled rapid turnaround times for support requests,” he says. “AI-based prediction and context management help expose precise discovery mechanisms through simpler interfaces like chats. Machine learning-based cognition engines make query resolution and policy support issues accurate and easy to deploy over secure channels like MS Teams and progressive web apps. ”

Investigate and push for the most impactful technologies. “Price optimization, predictive maintenance, and conversational AI technologies have the most impact, as the data needed to train them and keep feeding them as they learn tends to be plentiful. “Wu explains.” Their deployment also does not require a major change in business operations. Additionally, as many vendors offer these solutions, the total cost of ownership is relatively low compared to the revenue impact these technologies can generate. Sheth sees the greatest potential in multilingual NLP, machine learning and predictive AI.

Guarantee AI fairness through greater transparency. To gain acceptance and support for AI across the enterprise, the results provided need to be as fair and unbiased as possible. “Transparency and fairness are essential to the success of AI because they generate trust by informing both employees and customers about how they are governed,” says Baldwin. “There are plenty of examples of AI not working properly, and as a business the last thing you want is to be unable to explain why an error has occurred.” Still, there is still a long way to go in this direction, says Wu. “Many industries that are starting to take advantage of AI are focusing more on how their AI works and on a positive ROI with data first. limited available to them. For these industries, fairness is not an immediate priority, although it is a regular part of company narratives. While everyone is talking about putting ethics and fairness in AI first, not everyone is taking subsequent steps to tackle prejudice. “

Encourage awareness and training of fair and actionable AI among IT managers and staff. IT managers and staff should also receive more training and awareness to mitigate AI bias, urges Sheth. “AI is as good as the data we give it. Since humans are responsible for training data, there is a good chance that our AI algorithms will be corrupted through human bias or reflect any other type of unfavorable pattern detected over time. We can determine various models that can help make better and fair decisions, but at the same time, business leaders need to be aware of these challenges and make the right decisions to help eliminate biases when it comes to data. ”

Encourage awareness and training for fair and actionable AI at all levels of the organization. AI can be a business, but IT managers can lead the way in making sure AI works as it should. “Training and educating IT managers and staff is a good start, but often not enough,” says Wu. “Mitigating AI biases should be everyone’s business, as should data security. because it is akin to the business ethics of a company. ”

At the same time, he adds, “employees often need to be encouraged to adopt new professional behaviors before they become second nature. These incentives do not always have to be monetary. For example, corporate gamification can be used to raise awareness and generate interest in AI bias mitigation. It can be a lever within a company to gamify awareness of the problem of AI biases, elicit positive behaviors that help identify these biases and even seek potential solutions.

Regular review of AI results is also mandatory for success, says Sheth. “In fact, it’s been one of the hard-learned lessons for AI companies to always have humans in the loop.” It recommends “regular reviews of the randomly selected IA results, ensuring that all strata are correctly represented in the random sampling. End users may not always have the time and inclination to provide feedback on sub-optimal AI results. Evaluator comments are automatically fed back to the next model training round. This practice prevents models from becoming obsolete and irrelevant. “

Source link

Comments are closed.