You have all watched movies depicting rogue artificial intelligence machines wreaking havoc in humans' lives and leaving us wondering if this is a feasible scenario. The answer to that question should similarly address the issue as to whether Artificial intelligence can also replace data scientists (and lots of other human input terminals) altogether, who also happen to be intertwined in the design and genesis of the same Artificial intelligence systems. Both the movie scenario and AI replacing data scientists will be the result of considerable advancements in the science of Artificial intelligence and related fields, which have exponentially advanced for decades.
Lately, we have been warned loudly by people who ought to know the stakes well that Artificial intelligence does pose potential threats on humanity across various fronts and asks to tread carefully as we develop these systems. The punchline is that AI, like fire, is a useful servant but can be a bad master. None other than physicist Stephen Hawking, Nobel Laureate Frank Wilczek, computer scientists Stuart J. Russel and Roman Yampolskiy amongst others, Bill Gates, Elon Musk, and many more came out voicing concerns about the probable perils of advancing Artificial Intelligence mindlessly.Sometimes it does appear like the people sounding alarm bells, who also happen to be highly placed in academia and related fields, are fear-mongering, but is that the case? Are certain professions at risk of either being rendered obsolete or replaced altogether by advancing Artificial intelligence?
The modern data scientists are a highly specialized scientific agent extracting sensible and coherent knowledge gleaned from inferences arrived at from analyzing data sets for purposes of comprehending real-life phenomena. He/she handles big data and sifts through it mapping trends and patterns in a bid to understand and make predictions based on the said patterns. Data science is conducted within a broad interdisciplinary field incorporating data collection and analysis, statistics, and machine learning within a framework that deciphers and seeks to understand phenomena using algorithms.
Artificial intelligence on the other hand, which this article seeks to discuss whether it can replace the data scientist, is an impersonal human-made device that takes in and deciphers certain stimuli (read data) from its environment and makes several optimized actions based on the data received to increase its chances of carrying out a specific task effectively. They (artificial intelligent creations) mirror human cognitive functions usually associated with the mind and thought, such as learning and problem-solving.
Owing to tremendous advancement in the science of artificial intelligence, only that which has not been achieved remains AI, whereas already executable functions are relegated to routine actions. This phenomenon is referred to as "AI Effect" and is aptly observed in Teslas' theorem, which quips "AI is whatever hasn't been done yet."
Artificial intelligence has developed into two major pathways over the years to fill niches generally brought on by demand in the industrial market, namely:
- NARROW AI: This is artificial intelligence normally operating within narrow constraints to perform a particularly simple task exceptionally accurately. The result is a near-perfect execution of duty. Generally, these have little room for learning or performing new tricks and thus remain in intelligence stasis. Even their mechanics, should they be so enabled, allow typically for few specific movements controlled like clockwork. Essentially nothing comes from out of the box.
- GENERAL AI: This is the kind of stuff imagined by movie scriptwriters and is the subject of this article. The most straightforward description we could give General Artificial intelligence is a machine with general (without constraints) intelligence that can use that intelligence to solve problems. This is an entity that can have autonomy in decision making based on the desired purpose of its mission and the data it receives. General intelligence means the ability to ask and tackle open-ended questions that are not constrained by simple linear algorithms. It very much can be referred to as 'thought.' What makes this even creepier is that these AI can be constructed in such a way that they reprogram themselves into better machines as they learn more and more of their surrounding environments.
So, will AI eventually replace data scientists as posed by the title of this page? Will it depose other human intellectual endeavors and professions? Despite the appreciation of the growth and advancements in algorithms running Artificial intelligence, many experts in the field have come out to allay fears of such an eventuality taking shape in the near future. There are considerable challenges in the way of AI attaining usefulness comparable to that of data scientists, or any other highly specialized general intelligence types of professions. The first drawback in AIs march towards general intelligence is the fragmentation of research leading to non-linear progress. It is spliced up into fields that fail to communicate with each other effectively and are based on individual considerations. For example, machine learning or robotics, logic or neural networks, and even philosophical and social differences have been known to cause rifts between fields.
To replace the data scientist, Artificial Intelligence should perform certain functions at the same level or better than humans. These can be listed as follows :
- Thought. strategy, problem-solving, making judgments against open-ended challenges.
- Represent knowledge including common sense knowledge
- Communicate in plain language
- And implement all these skills towards a common outcome.
- It would also need to detect and respond to danger or hazard.
So far, these collective abilities, which amount to general intelligence, which is necessary for AI to replace the data scientist, are far from being realized in one individual. For these abilities to emerge, the AI will need to have brainpower with the capacity to process data along the same lines as humans. This is no small feat, and modern AI is still far from attaining this overall computational ability. The human brain has complexities that humble even the most seasoned neuroscientist. Its operation remains largely mysterious, and replicating one is still the stuff that makes movie scripts in the sci-fi world. Some interfaces join the human brain with machines being developed today, but that does not offer an autonomy of thought to the gadget being fused with the human brain.
We have it on the experts' authority that it will be a while before machines can generate and process thought by themselves. In the meantime, we will consider them, and they will process our data.