As I begin to write an essay on the applications of artificial intelligence (AI) in business, I took a moment to reflect on the potential of AI, as well as the common misunderstanding surrounding this technology.
My initial encounter with ChatGPT occurred shortly before I joined the diploma. I expected an ordinary and buggy chatbot. What I experienced was far from that. After only a few minutes of conversation, I was struck by the realisation that history was unfolding before my very eyes.
While technological breakthroughs such as the internet, personal computers, and mobile phones had transformed the world, I had never before experienced such a clear and significant moment where technology was advancing rapidly to an unprecedented degree. The whole experience was stunning to the point that I looked at the calendar and began to wonder if this is the date I should write down to mark the beginning of a completely new era.
AI has been changing the way we eat, live, teach, study, and consume information for some time now, but it has exploded in popularity in recent years to the point where one can hardly open a newspaper without seeing an article about AI.
As I interact with like-minded professionals and professors, I have come to appreciate the importance of practical AI implementation in day-to-day business practices. When discussing AI and machine learning with friends and family, I discovered that there are many misconceptions surrounding the field, and I felt compelled to address them in this article.
Myth 1: it’s difficult
One of the most prevalent misconceptions is that AI and machine learning (ML) are difficult to understand and learn. As someone who initially thought of Python as a snake, as opposed to the programming language, I can certainly relate. However, the use of no-code solutions such as Create ML, Amazon Sagemaker, AutoML, and Lobe has made it possible to create AI-infused solutions without having to write complex, technical code.
While machine learning can seem intimidating, it is essentially statistics on steroids. With a good understanding of business problems to be solved with this technology, a data-savvy person can begin incorporating machine learning into their work.
Myth 2: AI, ML, DL are all the same thing
I have seen too often, people referring to AI, machine learning, and deep learning (DL) interchangeably as if they are all the same thing. While definitions of AI vary, for the purpose of this article, AI refers to a broad set of tasks that computers perform to mimic human behaviour. Machine learning is a subset of AI that allows computers to learn based on the data they are fed. It has the ability to discover data patterns without requiring specific rules to be set by the programmer. Deep learning, meanwhile, is a machine learning technique that has become popular with the development of tools such as ChatGPT and DALL-E.
Myth 3: AI will take our jobs
While it is true that AI can outperform humans in certain tasks, there is no need to panic. Machines infused with AI are still a far cry from perfect, and they struggle with tasks that come easily to humans. If you’ve ever seen an AI-injected robot trying to pour a glass of water, an activity that’s a no-brainer for a four-year-old, you understand what I mean.
As AI continues to develop, while some jobs may be lost to automation, other opportunities will open up, freeing individuals from tedious and repetitive tasks. Properly harnessed, AI has the power to liberate us from mundane tasks, while simultaneously enhancing overall productivity. AI can be a valuable asset for professionals, as it can streamline activities such as scheduling meetings and optimising project timelines.
Speaking from the point of view of my profession (program and project management), by relying on AI to perform these tasks, we can instead focus on activities that require human intuition and interpersonal communication, such as stakeholder management and decision-making. In short, love it or fear it, AI will become your new co-worker.
The ongoing advancement of AI is generating considerable discussions on the subject of possible job displacement arising from increased automation. It is imperative to continue the debate concerning the prospects of these technologies and explore policy interventions that could help alleviate their adverse outcomes. For further insights, listen to the ‘AI and white-collar jobs’ recording from Davos 2023.
Myth 4: AI will become alive and destroy the world
While AI technology is incredibly powerful, it is not programmed to take over the world and has no appetite to cause global destruction. It is essential to understand the distinction between narrow AI, which is currently in use, and theoretical concepts such as general AI and superintelligence. The latter refers to the likes of Hollywood’s Terminator and Star Wars, or Asimov’s machines, which are not yet possible.
Myth 5: AI is unfair
There is a common misconception that ML models have built-in biases and are unfair. It is true that AI can produce biased results. The source of bias is not the AI itself, but rather the dataset it is trained on and the humans who decide what information to feed the machine. If we prioritise responsible AI guidelines, we can address these concerns. An interesting concept we explored is using the Hippocratic Oath, which doctors pledge, to guide the AI ethics.
Even if we are not regulators, Big Tech programmers, distinguished professors, or politicians, we can take action to make AI more equitable. As consumers, we can for example ask about the methodology used by the AI tool we’re using and raise challenges when we spot something not looking right. Additionally, the explainability of the ML models (so called glass-box) is crucial.
Myth 6: AI is expensive
While training a complex ML algorithm can be expensive, it does not always need to be. There are many open-source tools and pre-trained ML models that lower the entry barriers for small and medium companies. Furthermore, due to the current ‘AI hype’, companies often try to implement AI or ML at any cost, without understanding the business needs. In some situations, an intelligent automation or a simple regression model would suffice.
Myth 7: it’s all about the algorithm
While algorithms are essential, they are not the only component of a successful AI solution. A helpful way of thinking is that successful AI solutions require four primary components: data, algorithm, graphics processing unit (GPU), and human intelligence. When one of the first three components is of poor quality, human talent can help solve various data problems through, eg, synthetic data. I would argue that understanding the business problem, data-driven decision-making mindset, data advocacy and literacy are more important than a 100% accurate algorithm itself.
Myth 8: AI will save us
While AI has great potential to help us achieve the UN Sustainable Development Goals more quickly, it is not a silver bullet solution. AI can help reduce inequality, solve challenges in healthcare and discover new drugs. AI can combat climate change by reducing electronic waste, preventing plastic pollution, using green cooling technologies in data centers, and using satellite imagery to prevent wildfires or fight deforestation. However, we must not forget that the environmental cost of training the model is still significant.
I’m firmly convinced that the deployment of AI can serve as an effective instrument for addressing some of the most pressing global challenges, while also potentially alleviating social disparities. However, such adoption must be coupled with a rigorous and deliberate approach to policy making, which can ensure equitable distribution of those benefits. It is our obligation as engineers, programmers and citizens to use technology as a force for good.
In conclusion, AI has the potential to transform the way we live, and it is not as difficult to learn as some may think. Implementing AI successfully depends on driving organisational leadership and changing mindsets, as well as having the basic technical understanding of the AI concepts. It is a dance between humans and machines, not a dance of one. For AI to help us tackle the ‘world’s to do list’ it is critical to comply with responsible AI guidance and strike a balance between black box and glass box approaches. With careful planning and preparation, AI can become a valuable tool to be embraced, rather than feared.
ChatGPT in action
Bonus: after writing this blog post, I asked ChatGPT about common AI myths. See below for yourself how the article would look like if AI had written it itself. This time, it provided accurate information, however, it is essential to verify its responses for inaccuracies as it has the tendency to lie and (what is possibly more worrying) lie convincingly.
Demystifying AI: a ChatGPT take
There are many myths surrounding artificial intelligence (AI), some of which include:
- AI will replace human workers entirely: one common myth is that AI will replace human workers entirely. While AI may automate certain tasks and processes, it is unlikely to completely replace the need for human workers.
- AI has human-like consciousness: another myth is that AI has human-like consciousness, which is not true. AI is simply a set of algorithms and computer programs designed to mimic human behavior.
- AI is infallible: AI is not infallible and can make mistakes, just like humans. AI algorithms are only as good as the data they are trained on, and if the data is biased or flawed, the AI will produce inaccurate results.
- AI will take over the world: this is a popular myth in science fiction, but it is unlikely to happen in reality. AI is created and controlled by humans and cannot act independently without human input.
- AI is only for large corporations: AI is becoming more accessible to businesses of all sizes, and there are many open-source AI tools and platforms available for anyone to use.
It is important to dispel these myths to better understand the capabilities and limitations of AI and to use it responsibly and effectively.
Spoiler alert: yes, the photo is also fully generated by AI.
