Data science is constantly changing; in fact, it continues to develop and evolve, playing a crucial role in all industries and driving technological advances. I know you probably have these crazy questions going through your mind;Do we still need data science, or are tools like ChatGPT taking over most of the tasks?
· Do we still need data science, or are tools like ChatGPT taking over most of the tasks?
· Now that artificial intelligence exists, is it still worth training your own machine learning models?
· Is it still worth learning Python now that artificial intelligence is here?
· Now that artificial intelligence is here, is there still a need for data science? Or are we going to declare it dead?
All these new AI capabilities sound a bit overwhelming and make us both worry and wonder what we should do. What's left for us to do, especially as data scientists, model trainers, Python programmers? You might say.
First of all, AI doesn't build things on its own, it doesn't train models, it doesn't write Python scripts, it doesn't create specific workflows; we need to explain how this happens. Let me give a good example. We all know Minard's iconic data visualization of Napoleon's 1812 campaign in Russia. Charles Joseph Minard's map of Napoleon's lost campaign in Russia in 1812 is probably one of the best charts ever created and is considered one of the most influential examples of data visualization in history. If we wanted to recreate this data visualization example on Knime's Python code generation engine with the help of K-AI, we would need to give it a series of new tasks and then keep refining it until the results are what we expect. So even when using AI as a support, the project owner needs to explain the whole process in the next steps: what will be created, how it will be created, what data will be used, etc.
Second, AI does not check for accuracy. AI provides a result. It is not part of their job to assess whether it is correct. AI still needs fact-checking by an expert user: like fact-checking data science and business soundness. For this we still need a skilled end-user who knows what needs to be achieved and how. In case the result is not correct or does not correspond to the guided task, the end user needs to either correct it with better guidance or manually add the missing parts. This leads us directly to the third point: Fine-tuning AI models. There is now an emerging trend to fine-tune AI models. For this, you absolutely need data scientists.
If the images generated by AI and the graphic designers continue in parallel, AI can generate all kinds of images. But only the graphic designer at the end can verify the image quality and help with improvements if necessary.
On the other hand, with the continuous advancement in technology and the use of artificial intelligence in our daily lives, many may be worried about redundancy. Some even talk about the death of data science. Many say that data science is an oversaturated field and that machine learning is replacing data science. With the massive use of tools like ChatGPT and their use in coding tasks and more, we will live to see if data science is dead or alive.
Today, we have more and more data that generates valuable insights that guide decisions. These insights cannot be generated by a computer and we need them for data science. Machine learning models can be built and data can be leveraged to find valuable insights, but what really matters is the need for data and what to do with it.
To understand what to do with the data, you need people. You need data scientists! So, what has changed?
AI will get better and better, especially for basic tasks, and there will be less need for pure practitioners. But we will still need professionals who know the data science process and its mathematical requirements, who know how to correct and direct AI efforts and how to interpret the results generated by AI. In practice, we are moving from building and training models and services to consuming and improving them.
Long story short, we still need data scientists. Nevertheless, the role will probably change in the next future. There will be more focus on algorithms and the data science process rather than programming. At this point, low-code tools will make the whole process even more accessible and faster to implement. We will need more general data scientists who are well-versed in algorithm math, communicate well and have the ability to guide and correct AI towards the desired outcome.
Data science is not dead, but it is definitely changing. The best data scientist will not be the one who can code faster, but the specialized person(s) who can better guide the assembly of the data science project, taking into account data integration, data quality, data history, machine learning algorithms, result interpretation and process correctness.
Will we be more generalist? We will probably need more generalists to work more on the process in the first phase of a data science project. But we will still need specialized data scientists to review and refine the AI output. Just like graphic designers, data scientists will benefit from faster implementation of solutions through AI, but they will still need to be careful about the quality of the solutions provided by AI.
Data science is far from dead; it is an important and dynamic field that is constantly evolving. With the increasing volume and complexity of data, advances in related technologies and the growing need for data-driven decision-making, I can guarantee that data science will continue to be a very important field of study and practice for the foreseeable future. May 24, 2024
Sincerely regards,
Comments