Busting The Myths Of Artificial Intelligence
Conversations on artificial intelligence (AI) range from the extremely progressive views on the possibilities it offers to the other side of the spectrum where the chatter is all humdrum and the fear of losing jobs to machines overtakes everything else. Well, the fact is, both perspectives are right and matter equally.
Like with every new technology, changes due to advances in AI are difficult to comprehend in entirety since there are more unknowns than knowns. I am of the firm belief that we are barely scratching the surface with AI and significantly far from reaching the singularity. Hence, the need of the hour is to not fear AI and instead focus on understanding its relevance for our organizations and the human race at large. On that note, I have compiled a few myths that I often face during my advisory and consulting sessions with global end-user organizations.
Myth 1: Organizations do not really understand AI and it is something that we speak at events to take flights of fancy.
Hate to say it, but that is just not true. As per a recent Greyhound survey titled, State of AI 2017, more than 53 percent large organizations globally (we interviewed 5000 in 50 countries) are either already using AI for a project, conducting a proof of concept (PoC), or else planning to launch an initiative in the next 12 months. So, in other words, organizations are aggressively exploring ways and means to use AI over the next 3–5 years. But then there is the other 47 percent who either remain unclear about the use-case for their organization or have inhibitions around data privacy, security, compliance among other concerns. Having said that, all of these organizations (2350 to be precise) confirmed a good understanding about AI in general.
Myth 2: AI will cause extensive unemployment and organizations will need lesser people in the years to come.
In my humble opinion, this is a gross misrepresentation of the impact of AI on organizations and structures. While it is fair to say that AI is helping organizations automate basic tasks of repeated nature, they are far (read decades) from reaching a level of maturity where AI (and robots) can replace humans in complete. The above-mentioned survey confirms much the same. As per the survey, only 12 percent organizations believe they will replace humans due to AI and robots. In fact, a significant 68 percent confirmed their intent to hire new people to manage increased complexity arising out of the use of AI. Hence my assertion that use of AI points to the need for newer skills, hence newer types of jobs and not the proposed lesser jobs.
Myth 3: Scale-out architecture (read cloud computing) is the best way to manage AI implementations.
This is something currently only the most technically-sound understand. But in my view, this is a critical stepping stone in the AI journey and hence needs to be well understood by most. Against popular perception, not all AI projects must use scale-out architectures based on central processing units (CPUs). Albeit those involving data from supervised learning are surely a good use-case for using scale-out architectures, the data from unsupervised and reinforced learning is mostly better managed on a distributed architecture using a combination of graphical processing units (GPUs) and tensor processing units (TPUs). In fact, in my 9 of 10 advisory sessions, I see global organizations struggle with processing speed (when using CPUs) and cost of execution at scale (when using GPUs) while running machine learning (ML) algorithms. Hence the need for user organizations to remain wary of that sales pitch that basis all of AI use-cases on cloud.
Irrespective of whether we fully understand AI or not, it is a force that needs to be reckoned with. As organizations take their unique journeys with AI, they will have their fair share of learnings and over time learn to use it for tactical advantages. It is the individuals here who need to sit up and take notice. Those who refuse to re-skill will surely fail to fit in the new world order and hence must be ready to miss the boat. In the end, the onus of being relevant in the era of AI is upon our own selves and not our employers.