von Kyrill Poelmans | Mrz 10, 2023
…here’s your very own AI dictionary
Do you know what is currently the ultimate buzzword? Right, it’s AI – artificial intelligence. Over the last few months, it’s been practically impossible to avoid the subject. This is mostly due to the introduction of ChatGPT – the most talked-about AI application that you can use to generate content in just seconds (yes, we published an article about this as well – you can read it here).
With AI come terms like text generation, summarization, lemmatization, speech tagging, and BERT. If you’re not working with AI a lot yet, it’s hard to keep track of these terms. So to guide you through AI and its buzzwords, we created a little AI dictionary for you here.
Dictionary of the most common AI terms
Text generation
There are several AI applications that you can use for text generation. These applications use artificial intelligence to create written content. You can use this content for blogs, social media, reports, instruction manuals, and much more.
Summarization
Summarization – or text summarization – is the process of condensing large bodies of text into their most important parts. You can use it for static, pre-existing texts. There are two summarization methods: extractive and abstractive.
The first extracts the most important sentences from the original text. The second generates summaries itself, which may contain sentences or words that are not in the original text.
Lemmatization
Lemmatization is the process of finding the base form (the lemma) of the related word in the dictionary. It doesn’t simply cut off inflections. Instead, it uses lexical knowledge bases to find the correct base form of words. For example, a lemmatization algorithm knows the word better is derived from the lemma good.
Search engines like Google, chatbots and virtual assistants use lemmatization to better understand the meaning of sentences.
Speech tagging
Speech tagging – or grammatical tagging – is the labeling of the words used in a text according to their word types. The word types are: nouns, adjectives, adverbs, verbs, prepositions, conjunctions, pronouns, and interjections.
BERT
Did you ever search for something on Google that can have multiple meanings? Something like “how to catch a cow fish.” Up until recently, you would receive search results focusing on cows. With BERT, the search engine now knows you mean cowfish.
BERT is an algorithm based on AI. Computers and software can use the algorithm to better understand the meaning of sentences and words by looking at the context and order of the words. In other words, BERT improves the language comprehension of Google. The words catch and fish alert the search engine that you’re not asking about cows, but about cowfish.
Where Textmetrics fits in
If we look at the above dictionary, you might wonder how we would describe Textmetrics. Let’s give that a go here.
Textmetrics
Textmetrics is smart AI writing software that you use to personalize your content and increase its quality. It’s a simple approach to a complicated task. The text optimizer on the platform for text improvement gives you suggestions for improving the quality of your content. This makes writing easier and more efficient. You’ll write understandable, high-quality, and gender-neutral content that appeals to your entire target group.
Do you want to read even more about AI, our platform and more? Check out our other articles here.
von Kyrill Poelmans | Feb 22, 2023
…And how does it work?
When we say machine learning. You say, ‘That has something to do with AI, right?’ And you’re correct! Don’t worry about not knowing the exact definition of machine learning. A lot of people have to look it up! But luckily we’re here to explain it to you. So, let’s get to the bottom of what machine learning is.
Machine learning is a form of artificial intelligence that teaches computers to learn from data and to get better through experience, so you don’t have to keep reprogramming them with improvements. It is about training algorithms to find patterns in data and to use this analysis to make the best decisions and predictions.
Machine learning is used a lot in everyday life – in our homes, our shopping carts, our entertainment media, and in health care.
The difference between AI and machine learning
Machine learning and AI are often named in the same sentence. But they are not quite the same thing. Machine learning is a form of AI. So, machine learning is always AI. But AI is not always machine learning. AI is the umbrella term, and machine learning is just one of the many things under that umbrella.
Deep learning and machine learning are also often confused with each other. Deep learning is a subcategory of machine learning models. It attempts to imitate the functioning of the human brain and is used for speech recognition, computer translations, and facial recognition. “Deep” here refers to the number of layers in the neural network.
How does machine learning work?
Machine learning consists of different models using different algorithm techniques. Algorithms for machine learning are designed to classify things, find patterns, predict results, and make informed decisions. Depending on the data and the desired result, you can opt for one of four different models:
- supervised learning
- unsupervised learning
- semi-supervised learning
- reinforcement learning
The difference between supervised and unsupervised learning
Let’s have a quick look at the difference between two of the four machine learning models:
- Supervised machine learning
In supervised machine learning, a data scientist serves as an intermediary who teaches the algorithm what conclusions to draw. The algorithm is trained using a pre-labeled dataset. You can compare this to how you might teach a kid the different kinds of fruits using pictures and naming them. Once the teaching period is over, the computer (or kid) knows the difference between an apple and a pear.
- Unsupervised machine learning
In unsupervised machine learning, the computer gets the hang of complex processes and identifying patterns without the guidance of humans. Training is done using unstructured data without labels. If we go back to our kids and fruits analogy, the kid would learn by recognizing colors and observing patterns, rather than through us showing them pictures of the fruit and telling them the names.
What can we use machine learning for?
We can use machine learning to predict, discover and detect. These skills are useful for a wide range of things. For example, you can use machine learning to:
- Predict what customers are likely to buy.
- Detect mistakes while working on written content and receive suggestions for improvements.
- Discover the best possible moment to publish a social media post based on data.
The second item on the list is how Textmetrics uses machine learning. Enter content in the text optimizer on our platform for text improvement, and it will give you suggestions for improving the quality of the text. This makes writing easier and more efficient. You’ll write understandable, high-quality, and gender-neutral content that appeals to your entire target group.
Want to read even more about AI, our platform and more? Check out our other articles here.
von Kyrill Poelmans | Feb 6, 2023
… why a combination of AI applications leads to the best results
Have you used it yet? AI-generated content. Content created by artificial intelligence. Since the introduction of ChatGPT, it’s been impossible not to think about what AI can do. And what it means – or can mean – for the content you create.
ChatGPT is currently the most talked about AI application. But there are similar applications out there that generate content. In seconds, these programs can write an introduction, summary, blog article, or rhyme. Sounds good, right?
But does this mean there is no more writing left for you to do? Not quite. But what it does mean is that your working life can be a lot easier. And the quality of the content you write can be even better.
AI-generated content by ChatGPT
ChatGPT is a chatbot. A computer program or robot that you can converse with. You can ask it for gift ideas when it’s someone’s birthday, or ask it to write content for you. A summary, an instruction, an introduction, or an entire story. It’s like a search engine, but it answers your questions like a human would. Plus, it can generate loads of content in just seconds. Pretty amazing, right?
ChatGPT is not fully mature yet. Its answers aren’t always correct. And the content it writes is like a lot of content we find online – generic. It’s great at producing articles like “Three ways to save money” or “Five ways to find your dream job.” Why is ChatGPT particularly good at creating this kind of content? Because it uses content “fed” to it during training.
ChatGPT is not capable of thinking for itself. How you formulate what you want it to write is very important – one wrong word, and you might end up with the opposite meaning than the one you want!
The difference between ChatGPT and Textmetrics
It’s pretty impressive what ChatGPT can do. But it’s not mature enough yet to use it to create all your content. Just ask it to write several articles on the same topic. You’ll notice that the results are all pretty similar. That’s where ChatGPT differs from Textmetrics. Both are AI applications, but you use them for different reasons:
- You use ChatGPT to generate a general piece of content in seconds.
- You use Textmetrics to personalize your content and increase its quality.
How to get the best results
Content generated by ChatGTP might not have the right tone of voice for your audience. And when you use it to write a job description, it probably won’t use the right job motivators. But that doesn’t mean it isn’t useful at all. It’s just not ready to use right after it’s been created. You’ll have to combine it with Textmetrics to get the best results.
You can use Textmetrics to personalize all content – whether AI-generated or written by a human. The text optimizer on our platform for text improvement gives you suggestions for improving the quality of your content. You’ll end up with unique, high-quality content that appeals to your entire target group.
von Kyrill Poelmans | Dez 8, 2022
What do you do with a text you have spent hours working on, but its message does not come across to the target group at all? Throw it in the bin and start again, or is there anything left to salvage? What if you learn to communicate well? Then at least this won’t happen to you in the future. That’s what communication models are for, right? Unfortunately, it is not that simple. Communication models don’t teach you how to communicate. It is too much of a challenge for that.
However, communication models do give you insight into the pitfalls of communication. An important first step towards better communication. After all, communicating well is extremely difficult. Emotions, cultural differences, bias and even the medium you choose… they all play a role in how your message comes across. Try taking all that into account.
The NLP communication model and the 5 axioms
Do you search for ‚communication models‘ on Google? Then you quickly see that there is no shortage of models. Don’t worry: we are really not going to discuss them all here. However, there are two we would like to highlight:
NLP communication model
Here the focus is on the senses. How your senses perceive a message determines how you react. With that, the model looks mainly at unconscious processes in the brain: the omission, distortion and generalisation of information. How you filter information determines how you think, and therefore how you communicate.
You do the filtering of information based on:
- Beliefs and values (which determine what information comes in).
- Time and space (your senses perceive information relatively).
- Language (do you understand what the other person is saying).
- Memories and the feelings you attach to them.
The 5 axioms
This communication model states that you are actually always communicating, even if you don’t do so consciously. You don’t even have to respond to give a reaction. Is the relationship between sender and receiver not good? Or is the message not coming across clearly? Then there is noise on the line. What the sender says comes across differently to the receiver. Noise is actually always there, but a little is no big deal. Therefore, the goal should always be to minimize noise.
Common communication pitfalls
Be aware of unconscious processes in your brain, know that you are always communicating and avoid noise… It’s quite a lot to consider when communicating. So what are these unconscious processes? And how do you avoid noise? First, consider bias. We often communicate based on what we know. That might be why your text mainly addresses men. Or people of your own age and ethnicity. Or there will be noise on the line because you write from your own expertise. For example, you use jargon that not everyone is familiar with.
These are common communication pitfalls that prevent your message from arriving as you intended. Understanding these pitfalls is a first step. But how do you avoid falling into them? Is there an easy way?
Most certainly! At Textmetrics, we have the solution for you. As text improvers, we know exactly which pitfalls there are. We developed the platform for text optimisation. A simple approach, for a challenging task. The platform gives you suggestions and advice that make writing more fun and more efficient. You write good, inclusive texts in understandable language, appealing to your entire target group.
von Kyrill Poelmans | Sep 2, 2022
Are your job ads free of bias? So much so that you succeed in building a diverse and inclusive workforce? Then you’re doing a great job. Unfortunately, though, many companies still struggle with this side of recruitment. They find it hard to hire people from different backgrounds, genders, and ages. The answer to this problem lies in a diversity recruitment strategy. A diversity recruitment strategy aims to eliminate bias from the hiring process. You want to recruit candidates in a more diverse and inclusive way. Inclusive hiring is an important first step towards a more diverse workforce. A workforce that reflects society and welcomes everyone.
How to kick-start your diversity recruitment strategy
It can be challenging to develop your diversity recruitment strategy. These three tips will help you on your way.
- Proactively reach out to underrepresented candidates
You can make the greatest difference in the first stages of the recruitment process. A more diverse workforce starts with hiring from a diverse pool of candidates. If these people don’t apply, you should actively look for them. You can search for people on LinkedIn, for example, and invite them to apply for your job.
- Reduce bias in the recruitment process
Gender and age bias often stand in the way of building a more diverse workforce. Older people, for example, get invited to interviews less often than their younger colleagues do. The same goes for women versus men. Although they should be treated the same, they are not. Men still have more chances of being invited for an interview and hired for a job. But everyone deserves an equal chance to be interviewed and hired!
- Tailor your message to the entire target group
Job descriptions are often not written to appeal to everyone in your target group. Bias plays a role here too. Without you even knowing, job ads regularly contain more masculine words. These words discourage women from applying. They are also frequently aimed at younger people, leaving older people feeling that there is no use in applying. These things stand in the way of becoming more diverse. You should try your best to tailor your message to your entire target group, without excluding anyone.
Use Textmetrics to kick-start your diversity recruitment strategy
You can use the Textmetrics platform to kick off your diversity recruitment strategy. Using algorithms based on artificial intelligence, it reads and analyzes your content. You’ll then receive real-time suggestions for improvements. These improvements will help you write jobs ads using more inclusive language, so they are free of bias. They’ll have the right tone of voice to appeal to everyone in your target group. Just what you need for a successful diversity recruitment strategy.
Want to know more about our Smart Writing Assistant? Click here.