5 Most Common Myths of Artificial Intelligence
MYTH #1 AI is going to replace all jobs
It is true that since the start of developing AI, it has replaced certain occupations and has the potential to seriously disrupt labour. However, seeing that AI is aiming to replace all jobs instantaneously of labour from humans to machines is a tremendous over-simplification.
There have been certain transformations of employment since nineteenth century and there has been a number of occupations since the rapid development of population which has generally been consistent. Regardless of what is being said, there is exceptionally little proof to propose any mass unemployment or widespread redundancy of human workforce’s is likely. It is just as possible that a more productive and beneficial economy can take place increasing the effectiveness and reduction of waste from automation promises, this in turn can grant more alternatives in investing time on productive and profitable pursuits.
This is to say that AI isn’t necessarily there to replace all humans from their jobs but to augment human workforce’s so they are able to work efficiently and find smarter and new ways - employers are using AI technology for this reason.
FACT #1 AI is going to augment all jobs
MYTH #2 AI is used only for billion dollar new problems
This is a common misconception of AI. Due to the reality that it is at the cutting edge of current innovation, it is easy to assume that only massive companies can afford to invest in it. But this is simply not true.
AI is also for those companies that you have used such as Siri, Alexa or indeed Google. One of the core imperatives of a business is to understand your clients. This was true in the first markets such as the agora in Ancient Greece and the forum in scientific Rome when buying and selling was done in person. This is still in effect and true today as many of the buying and selling is done on the Internet. Numerous companies utilise AI for this reason as these companies are sitting on a chain of unstructured data with emails or comments on social media. AI can categorise each individual thread and understand sentiment in comments which is beneficial to firms.
What’s more is that implementing AI is a lot easier than most individuals think. Lots of platforms and programs exist which can be easily overlaid on top of any existing processes and systems that can inform a greater understanding than data that you already have. AI is not only for billion dollar problems such as investing in driverless cars but can also be utilised in understanding your clients’ better and categorising email threads.
“Whilst AI is being used in industrial settings to model thousands of datasets, it doesn’t remove the need for data scientists – rather it limits and transitions these experts to deal explicitly with outlying and unusual cases.”
FACT #2 AI is used for million dollar existing problems
MYTH #3 Machines are better than Humans
For the last 30 years the media has loved to depict AI as meaning machines are superior than humans whether it be Schwarzenegger in the Terminator or Alicia Vikander in Ex Machina. Another case is Google’s DeepMind / AlphaGo triumph over Lee Sedol which was displayed as Machine defeats Human but this is essentially not true, rather it would be Machines plus numerous Humans defeats single Human.
On the left hand side is the Standard Model Lagrangian equation and on the right hand side is a picture of leopard print items. Machines are better at efficiently and swiftly detecting the solution to the equation whereas they might not do so great in finding the leopard print belt. Similarly, humans are much better at identifying the leopard print belt than they are in figuring out the solution to the Standard Model Lagrangian equation.
Machines and Humans have diverse capabilities and they complement each other. They are ideal in handling structured computation whereas humans are better at discerning meaning and context.
“We can teach a computer to recognise a car, but we can’t ask that same computer. ‘How many wheels does that car have?’ Or, ‘what kind of engine does it have?’ ”
FACT #3 Machines complement Humans
MYTH #4 Algorithms are more important than data
There is standard media scope of AI which is biased towards the focus on the Machine Learning algorithms as the foremost critical component. The stories about Machines defeating Humans in Chess and Go are examples of this media treatment. The media is centred on ‘deep neural networks’ and ‘deep learning’ and how the machines make decisions.
This coverage underlines and gives the impression that for a company to have a chance in applying for AI, they ought to contract machine learning specialists to construct the perfect algorithm for them. But if a business did this without considering the results as how they would amplify the high quality, high volume customised training data from which the machine learning model could learn, there would be a mismatch between expectations (we had the greatest specialists construct our algorithm) and outcome (our model is only 50% accurate).
Buying commercial Machine Learning service offerings nowadays from Google, Microsoft or indeed Amazon without having arranged a plan or budget for training data is much like buying a phone without access to the internet. You just bought a costly block of metal and glass. This example breaks down as the machine models get better the more you refuel it with more and more training data. This would be like a phone getting more minutes, texts and data after each month they pay for data.
This implies that quality and quantity of training data is just as critical as the algorithm that you use.
FACT #4 Algorithms are just as important as data
MYTH #5 AI can be completely objective and bias free
It would make logical sense that a computer would compute without inclination as after all, algorithms don’t have sentiments or opinions, but this is not the case. Artificial intelligence is created and informed by very biased creatures – humans! This is typically one of the biggest challenges that comes with implementing AI.
Human administrators majorly impact the processing and incorporation of power in AI by including values that AI ought to have and whether the algorithms we code should or shouldn’t make correlations.
Although AI can never be completely objective, with the correct contemplation, it can get close to being that way.
FACT #5 AI can be completely subjective and biased as the humans who built them
You can replace the 5 Myths of AI with the 5 Facts of AI
Fact 1: AI is going to augment all jobs
Fact 2: AI is used for million dollar existing problems
Fact 3: Machines complement Humans
Fact 4: Algorithms are just as important as data
Fact 5: AI can be completely subjective and biased as the humans who built them