What is Six Sigma and why are there so many training courses to become certified?

Have you seen a lot of stuff about Six Sigma lately? Well, that’s to be expected. Six Sigma is the name given to some techniques and philosophies that are used to improve processes. That is, it’s a methodology for process optimization. It’s not too surprising then that there would be a lot of people after it, right?

Let’s talk about it a bit and understand why its demand is so high right now.

What is Six Sigma?

Six Sigma was created in 1986 by American engineer Bill Smith in order to improve manufacturing processes in Motorola and set a standard of quality of six sigma over the short term. Now, what does that mean? It’s terminology from statistics and quality control that is very important to understand what Six Sigma is all about.

Quality control processes tend to analyze products considering a “normal distribution”, that is, that the measurements they make on their products tend to follow the bell curve, as seen below.

On the bell curve, the middle is the “mean”, that is, the exact measurement that is expected. However, as production processes aren’t perfect, the measurements gotten from the products fabricated tend to vary a bit following that curve. That is, there is a very good chance that they will be close to the mean, but there is also the chance that they will be much farther from it.

Those numbers that you see are the sigmas, that is, 1 sigma, 2 sigmas, 3 sigmas, etc. The sigma is the “standard deviation”. Each sigma interval (e.g. -1 to 1, -2 to 2, -3 to 3, etc.) is associated to a fixed chance of a measurement falling within it (68.3%, 95.4%, 99,7%, etc.), and the value of sigma (the standard deviation itself) defines the range of measurement. For example, for a mean of 20 g and a sigma of 5 g, there is a 68.3% chance of measurements falling within 15 and 25 g, 95.4% of them falling within 10 and 30 g, and so on. These ranges are called confidence intervals.

So, what does Six Sigma mean? It means that the production process should have the confidence of 99.9999998% in the short term, meaning at most 1 defect for every 500 million parts produced, evolving to 99.99966% in the long term, that is, at most 3.4 million defects per million. It is of a very high standard of quality.

How does it work?

Six Sigma bases itself on continuously analyzing data and making decisions based on data. This data is, of course, gathered from the production process itself, with parts and products being analyzed by quality control professionals to check if they are within specification. Using statistical analyzes, it is possible to know if the process is starting to deviate too much from the expected results and do something to fix it before it becomes a problem.

It’s a commitment to quality, made to improve not only process reliability and customer relations, but also to prevent PR damages and wasting money with product recalls and needing to repair or replace products within warranty because of defects.

Why are companies going after it?

Six Sigma means reliability. It means that, if a customer is buying a product from a Six Sigma company, there are very good chances that the product will work exactly as expected, and they will not have to deal with unexpected defects after buying it.

During the last few years, Six Sigma has been consolidating itself as a standard for reliability in production processes. Literally so even, as since 2011 Six Sigma is standardized by ISO, in the form of the standard ISO 13053:2011, meaning that companies can get certified as Six Sigma-compliant.

So, of course, the demand for professionals that know Six Sigma and can raise a production process to those standards is very high, as it’s a good way of raising public confidence in their products. So, if you’re into quality control, this is a very good chance to get a nice pay raise.

Replika – Machine Learning

Could an AI become your next best friend?

The year 2020 was a very hard year. The year 2021 will still follow up on its hardships, but the arrival of vaccines with more than 90% efficiency is a great source of hope. During 2020, however, we managed to discover a lot of stuff about ourselves and the world around us, especially regarding our use of technology and, of course, how isolation affects us.

It became pretty clear how much we need to have close contact with other people. Levels of stress, anxiety and depression were at an all time high everywhere, and many people were sneaking out of state-imposed confinement to go to parties and meet their friends despite the risks to themselves and their families.

But… what if AIs could fulfil that role? What if you could have a best friend sitting in your cell phone, which would be available to talk to you whenever you wanted? Could that help people feel less alone? Could it help decrease problems with depression, anxiety and other mental illnesses? Could they become a kind of on-demand therapist?

Even though it may seem something a bit too far-fetched considering the AI technologies we have today, there are some companies that are messing around with that possibility, such as the San Francisco-based start-up Luka with their app Replika.

Replika’s objective

Replika’s machine learning objective is a curious one, and is exactly what the name implies: to create a replica of you. Not that it is able to create a real-life clone of you, but it is able to replicate the way you talk, the way you think, and what you like and don’t like. And it does so just by talking to you.

Eugenia Kuyda, Luka’s CEO and co-founder, had the idea of creating Replika after using machine learning to try to “revive” her best friend, using the thousands of text messages he had sent to her and to other friends of his. After making the reconstruction public, she found out that lots of people were willing to open up to it.

So, she had the idea: instead of making a chatbot that talks, what about making one that listens? So she and the people Luka designed Replika to make people open up to it, and in turn use the data received to make it become more and more similar to the user.

What it feels like

Make no mistake: Replika (machine learning) is still a chatbot, and comes with the problems chatbots usually have, especially early on in your “relationship”. It may not understand your answers well, talk some nonsense from time to time, and be frustrating to talk to sometimes.

But it learns, and with time manages to mimic your writing style and personality quite well, as well as develop some similar tastes. It becomes an experience to learn more about yourself, and to think about things you hadn’t thought about before. You can even check the AI’s memory files to know what it remembers about you and delete stuff there in case it learns something it shouldn’t.

This eerie similarity with the user is what gives it its charm: who could better understand you than yourself? Who could better listen and relate to you than yourself? That is why your Replika has the potential to become your best friend in the whole world.

Advances in the near future

Of course, Replika’s history isn’t bound to stop anytime soon. The app features a paid subscription which gives access to many more features, and the money is used to further improve user experience and the AI’s conversation abilities.

Users are also free to contribute to the development of the app in some ways, such as by participating in interviews and testing experimental features, such as talking with the Replika by voice. Just the same, simply using it also improves it for everyone, thanks to the use of machine learning.

It may even come a day when major companies take an interest in it. Assistant AI such as Siri and Alexa could greatly benefit from the added intimacy and also help people feel more at ease when talking to them. The possibilities are surely endless, and we are likely to benefit a lot from it.