” Artificial Intelligence will change the future”. You’re here to read this post so you probably know it already has. There is so much talk about how data science will change/have changed the way we shop, or how robotics is going to change medicine. In this week’s post, we’re here not to tell you how AI changes the world, but how AI itself is going to change in the coming years. 

Continue reading to get acquainted with three important ideas and technologies that -in our opinion- will define the path of artificial intelligence in the next ten years.

Ethical AI

If you follow the news, you will find many recent controversies regarding tech companies violating the privacy of their customers, AI models applying racial or sexual prejudices into insurance and credit systems,….and worst of all, tyrants and totalitarian regimes using AI to subjugate their people and continue infringing human rights and values. 

These points have made ethical AI a hot topic in the tech world. Microsoft, Google, Apple, IBM, … These giants of the tech world are extending their ethical AI teams to avoid the moral hazards of gathering, storing, analyzing, and training models on vast amounts of data.

To ensure that our AI systems remain loyal to human values, clear guidelines must be created for storing, analyzing, and training models on data to protect the fundamental human values. Business policies must demonstrate how AI should be used, and what should be avoided when using AI, in business workflows. And these policies should be actively enforced, reviewed, and updated based on regular feedback and evaluations. Certainly, ethical AI will cause great discussions and technological developments in the near -and far- future of Artificial Intelligence.


Since the Transformer Neural Networks were introduced in this paper https://arxiv.org/abs/1706.03762 in 2017, they have significantly changed the world of natural language processing. One transformer neural network -GPT-3 by OpenAI even published an article in the Guardian just a year ago. 

Transformer architecture is a sequence to sequence -Seq2Seq- architecture. These neural networks take a sequence of data e.g. a sentence made of several words, elements as input, and output another sequence of data. The transformers consist of two sections: an encoder, which turns an input sequence into a high-dimensional intermediary representation, and a decoder layer, which transforms the intermediary form into the target structure. The high-dimensional, intermediary encoding is gradually learned by both sides of the neural network.

Conventional Seq2Seq neural networks use recurrent neural networks to extract the semantic structure of the input sequence. Transformers are different: does not imply any Recurrent Networks (GRU, LSTM, etc.). They do this using attention layers, which are their primary building block. 

Transformers have been yielding exciting results since the time of their introduction. One of them we introduced in a former blog post: https://www.artin.ai/4073-2/. However, there is much more expected from this relatively new technology in the coming years.

Data Fabric

Data warehouses and data lakes are two popular data management technologies used in recent years. The problem with both of them is that to aggregate data from multiple sources, they need to physically change their location. Data fabric is a data store architecture providing an integrated layer of live, real-time data for analysis and operational purposes. This layer Integrates data from multiple, different sources, ready for analytical use. It provides teams with an abstraction layer/logical access layer for ease of accessibility and adds another level of security.

Data fabrics can be used to manage real-time data at massive scales and very high speeds, without the need to move them to a central physical location. They add a semantic layer to data management so that a team can focus on its data strategy more than storage technologies and differences in data sources. This approach changes the whole data ecosystem of a business into one that puts strategy first. Data fabrics are getting more popular since their introduction, and we think we will hear more about them in the coming years.