Saturday, April 13, 2024
HomeSECURITYnot a magic wand for performance: why you need to be realistic

not a magic wand for performance: why you need to be realistic


AI Is Not a Magic Wand for Productivity: Why You Need to Be Realistic

How to Use Artificial Intelligence Effectively and Avoid Common Mistakes

Many developed countries are facing the lowest productivity growth in 60 years. So it’s no surprise that some see artificial intelligence (AI) as a productivity savior. The media is announcing a new era of high performance powered by AI, especially generative AI tools such as ChatGPT and DALL-E. Leading scientific journals are full of examples of how AI has enabled breakthrough discoveries in research. Machine learning has been used, for example, to predict the shape of proteins from DNA information, or to control the shape of superheated plasma in a nuclear fusion reaction. One team at CSIRO has developed an AI-based autonomous system that can produce and test 12,000 solar cell designs in 24 hours.

Does that mean we can flip the switch, leave it on automatic, and head to the beach? Not really.

Not a performance panacea

The examples above are encouraging, but also a distraction from the many AI applications that didn’t work. These are the cases where the use of AI was costly and time consuming and did not lead to the desired result. They are often not recorded in magazines and media.

In 2021, the AI ​​community had to pause when 62 published studies that used machine learning to diagnose COVID-19 from chest x-rays were found to be unreliable and unsuitable for clinical use, mainly due to input data issues. It was a poignant reminder that AI is wrong.

That’s not to say that AI can’t be used to improve productivity – it’s just that it’s not a ready-made solution to our productivity problems. AI cannot magically fix problems associated with inefficient processes, poor governance, and negative culture. If you throw advanced AI into a stupid organization, the artificial brain will not make the company smart. AI will simply help an organization do stupid things more efficiently (i.e. faster). This is unlikely to improve performance.

Where AI applications work

One recent study by the US National Bureau of Economic Research found that the productivity of help desk employees who used an AI tool to help with conversations increased by 14%. In Australia, Westpac claims that AI has delivered a 46% increase in programmer productivity without sacrificing quality of work.

In many ways, these examples are not surprising. Clearly, AI can improve productivity if used effectively; Google Maps is clearly better than the old road atlas when it comes to getting from point A to point B. So what do situations in which AI work well have in common?

Successful AI applications are typically characterized by a clear need and function for the AI ​​system. They are well integrated into the broader processes of the business or organization and do not interfere with the performance of other tasks by employees.

AI Application Failures: Causes and Ways to Overcome

Achieving AI-enabled productivity benefits across an entire organization, let alone the entire economy, is difficult. Many organizations are still struggling with an easier digital transformation.

Consulting firm Deloitte estimates that 70% of organizations’ digital transformation efforts fail. Perhaps the real solution to the productivity dilemma lies not in the use of AI, but in the elimination of organizational inefficiencies that arise when new technologies are introduced.

Modern offices are often overwhelmed with useless emails, redundant meetings, and bureaucracy that saps the energy and motivation of employees. Routine work and distractions reduce their productivity.

It is unlikely that AI will be able to solve this problem. In our time, attention is a valuable resource; AI designed to protect us from unnecessary routine can become intrusive. We may even find ourselves in a situation where AI tools compete for our attention.

Economist Stuart Mills of the University of Leeds suggests that tools like ChatGPT, which automate bureaucratic processes, don’t affect productivity.

We asked a senior manager at an international engineering company if he uses ChatGPT. “Oh yes,” he exclaimed, “I use it to create all these reports that management keeps asking for. I’m sure no one reads them, so they don’t have to be high quality.”

The path to long-term productivity growth

It is likely that in the long term, AI will improve productivity at the societal level by bringing about transformative change.

As of September 2022, 5.7% of all scientific articles were on AI, up from 3.1% in 2017 and 1.2% in 2000.

Innovators around the world are exploring the possibilities of AI to accelerate their work and discovery. We can expect the most effective solutions that solve real problems to stand out and take their place.

The success of AI implementation requires understanding the context of the application of the technology, choosing the right tool for a particular task, and using it correctly. In addition, issues of process, management, culture and ethics need to be addressed.

Source link


Please enter your comment!
Please enter your name here

Most Popular