Artificial selection

How to make business comfortable with artificial intelligence

Entrepreneurs and pundits on the front lines of the AI ​​revolution recognize that there are issues such as a company’s culture or lack of customer trust that cannot be solved with technology alone. These are fostered by principles that shape the day-to-day internal and external functioning of a business.

AI is a powerful mechanism for amplifying human knowledge, skills and efficiency. But can AI supporters use AI to correct a moribund or toxic corporate culture? This is probably the thorniest challenge with AI deployments.

One of the challenges that artificial intelligence models are based on historical data, which means that they are subject to biases that are embedded in the algorithm during their learning phase. So basically humans are passing their biases on to AI applications. Sometimes an automated process doesn’t take into account the people it works with.

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

Also read: Organizations need to take a multilingual approach to knowledge management capabilities

Extend AI Ownership and Engagement Beyond IT

The adoption of AI in a business should be a comprehensive business plan, with all parties involved. The successful distribution and production of AI is an effort for a variety of activities away from data science. Large teams need to move from the technical side, involving IT and cloud operations with data security and control, to the business side, which includes change management, instructional training, acquisition, and best practices.

AI does not seek to replace human beings, but enables human dialogue with the power of automation and ingenuity that a machine can have.

Identify AI in the areas where it has the most impact

The business components of promoting and delivering AI are very different from industry to industry, Wu said. But the common theme is that organizations need to have a reliable source of clean data. and wealthy as a product of normal business operations. Companies with large support centers often keep a good record of events and decisions. The activity data of commercial organizations is generally as precise as necessary for good accounting practices. This data will continue to add fuel to their AI / ML as it is read. On the other hand, while marketing organizations have a lot of data too, it is often noisy and often needs to be cleaned up before it can be used in AI and ML production.

Encourage awareness and training in fair and efficient AI between IT managers and staff

IT managers and staff should also receive additional training and awareness to reduce AI bias. AI is as good as the data we provide, because learning comes to them full of human biases. There are different models that can help make better and fairer decisions, but at the same time, business leaders need to be aware of these challenges and make the right decisions to help eliminate data bias.

Also read: IT Leadership: Five Effective Strategies For Coaching Rising Stars

Encourage awareness and training in fair and effective AI at all levels of the organization

IT managers can lead the way in ensuring that AI is effective. Reducing AI bias should be everyone’s responsibility when it comes to data security, just as it does with corporate business principles.

At the same time, employees often need some motivation to encourage them to adopt a new professional behavior before it becomes second nature. These benefits don’t always have to be financially sound. Commercial gamification can be used to raise awareness and generate interest in AI bias reduction. This can raise awareness of the problem of AI biases; elicit positive behaviors that help identify these biases and seek potential solutions.

It is recommended that regular reviews of randomly selected IA results be carried out, 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 suboptimal AI results. Actively and regularly evaluate the performance of your models. Evaluator comments are automatically fed back to the next model training cycle. This practice prevents models from becoming unnecessary.

Discover the new Enterprisetalk podcast. For more updates like this, follow us on Google News Enterprisetalk News.