Artificial intelligence

A young woman and man sit with serious expressions in front of a laptop.

Artificial intelligence or AI refers to a machine or a software that tries to imitate the intelligence of a human being and is capable of performing functions considered intelligent, such as learning and planning.

European Union’s Artificial Intelligence Act

  • The European Union’s Artificial Intelligence Act came into force in August 2024.
  • The act is the world’s first comprehensive regulatory framework covering artificial intelligence.
  • The act applies to both the development of artificial intelligence and its use after publication.
  • The act aims to also ensure that artificial intelligence is transparent, traceable, non-discriminatory and environmentally friendly.

Some artificial intelligence applications have been in use for over 50 years already. Artificial intelligence is a central part of the digital transformation of the society.

Artificial intelligence in everyday life

We don’t always even necessarily realise that we are using artificial intelligence. Below you can find some examples of everyday applications using artificial intelligence.

Online shopping and ads

  • Artificial intelligence is widely used to offer targeted recommendations. Recommendations are based on, for example, people’s previous searches, purchases or other online behaviour. In the retail sector, artificial intelligence is widely used in areas such as product optimisation, inventory planning and logistics.

Search engines

  • Online search engines learn how to offer users relevant search results based on the enormous amount of data that is constantly fed into them.

Digital assistants

  • Smart phones use artificial intelligence in order to make the product as individualised as possible. Virtual assistants answer questions, offer recommendations and help maintain daily routines.

Machine translations

  • Translation software, whether based on written or spoken language, uses artificial intelligence to produce and improve translations. The same applies to other similar applications, such as automatic subtitling.

Smart homes, cities and infrastructure

  • Smart thermostats learn to conserve energy based on our behaviour, and the developers of smart cities aim to improve transport connections and reduce congestion through traffic management.

Cars

  • Cars take advantage of security features based on artificial intelligence.

Cybersecurity

  • Artificial intelligence systems can help with recognising and blocking cyber attacks and other cyber threats with the help of new data, model recognition and analysis of completed attacks.

The fight against disinformation

  • Some artificial intelligence applications are able to recognise fake news and disinformation by analysing social media content and looking for sensationalist or otherwise alarming words as well as recognising official sources online.

Source: European Parliament

A young man wearing a cap sits outdoors, squinting one eye and looking straight ahead.

What is a deepfake?

Deepfake is a forgery created using artificial intelligence, where it is it possible to have a real person say or do things that they have not said or done in real life, often by combining images, videos and sounds.

Spot a fake

The rapid development of artificial intelligence has made it difficult to detect deepfakes. However, there are certain technical details that can help you to determine whether an image, video or audio recording is fake.

Images

  • Check whether the shadows are in the right place in relation to the direction of the sunlight, or whether there are natural reflections in windows or eyeglasses, for example.
  • If the image is sharp otherwise, check whether there is any blurring in unexpected places.
  • If the video was shot outdoors, check whether the wind is bending the plants naturally, for example.

Videos

  • Deepfake videos often replace a person’s face. Pay particular attention to facial expressions and the movements of the eyes and lips in relation to other parts of the face.
  • Can you see any blurring in the movement? Does the movement of the lips match the audio?
  • When a person is speaking, their tone of voice and speech rhythm vary. Listen to whether the voice of the person seen or heard is robotic and monotonous.
  • Is the skin on the face unusually smooth, exceptionally shiny or matt?
  • And are the movements and gestures of the person authentic in relation to the activities on the video?

Audio

  • In order to spot audio forgeries, you can ask the person something only they know.
  • Audio forgeries may also use some uncharacteristic expressions or ask you to do something surprising, such as make large money transfers.
  • You may also spot a scam if the person’s manner of speaking, speech rhythm and expressions sound unfamiliar.

Source: Kyberturvallisuuskeskus.fi

What are the challenges of generative language models that fall in the field of artificial intelligence?

Generative language models, such as Chat GPT, are an example of artificial intelligence. These types of applications produce text based on questions asked by the user. They are also capable of improving their answers, for example, if the user asks the application to provide more information on a particular topic or explains why they think the answer is not useful.

  • However, the Chat GPT application is not intelligent. It is not capable of independent thinking, so to speak.
  • The answers it provides are based on a language model that predicts what word follows another and which words often appear together.
  • Chat GPT can quickly produce text that can often even make sense, but the user has to stay alert because the answers may contain serious errors, outdated information and complete nonsense that the application has hallucinated.
  • Make sure to always check the facts provided by applications and software built on language models!

Be careful with AI-based image generators!

You need to know how to use image generators

Image generators create visual elements and images based on the user’s prompts. The generators go through large image databases and use them to learn about image structures and features and how different elements are usually related to each other.

  • In order to create an image that makes sense, you need to give the software good textual prompts.
  • In spite of this, the images often look quite ridiculous because they have been created by a machine.
  • A machine won’t understand, for example, why an arm can’t grow out of the stomach of a human figure, or why a human figure cannot be absorbed into the parquet flooring, for instance.
  • You should check the details in an image you created before you use it in an official presentation for work, for example.