InfrastructureΚυρίαρχη υπολογιστική ισχύς
On this site, computing capacity whose access, operation and data handling are under effective Cypriot control. Ownership, location, legal jurisdiction and dependence on foreign suppliers are separate questions, so a server located in Cyprus is only one part of the picture.
ExampleA government department choosing a cloud service looks at who controls access, where the data is processed and whether the service would keep running if a foreign supplier pulled out.
Regulation and governanceΠρόσβαση σε δεδομένα με κανόνες
Rules and controls that set out who may use data, for what purpose, for how long and with which safeguards. Data held by public bodies is not automatically open, nor automatically available for training AI models.
ExampleAn approved researcher studying hospital waiting times works with a limited dataset in a secure environment. Each access is logged, and access ends when the project finishes.
Regulation and governanceΔιασφάλιση αξιοπιστίας της ΤΝ
Checking claims about an AI system’s performance, risks and controls against evidence, before the system is put into use and while it is in use. Review by an independent party adds confidence in the results.
ExampleBefore a municipality launches an online assistant for residents, a reviewer tests whether it invents requirements, records each failure and repeats the tests after every update.
Regulation and governanceΡυθμιστικό δοκιμαστήριο
A supervised, time-limited arrangement in which an organisation that meets the entry conditions tests an innovation with a regulator under agreed terms. Taking part does not remove legal duties or automatically authorise a public launch.
ExampleA financial technology company in Limassol checks whether its planned test falls within the scope, eligibility and authorisation conditions of the CySEC sandbox.
Regulation and governanceΤΝ υψηλού αντικτύπου / υψηλού κινδύνου
High-impact is a general description of uses with significant consequences for people. High-risk is a specific classification under the EU AI Act, based on a system’s intended use and legal criteria, with exceptions.
ExampleA bank’s system for assessing whether a loan applicant is creditworthy is treated differently under the AI Act from its system for detecting financial fraud, although both are used in banking.
Regulation and governanceΚανονισμός της ΕΕ για την ΤΝ
The EU regulation on artificial intelligence. It prohibits certain practices and sets different requirements for particular systems, uses and general-purpose models. Which duties apply, and from when, depends on the category of system and the organisation’s role.
ExampleA software company that develops a CV-screening tool and a hotel group that uses it to sort job applications can have different responsibilities under the Act.
InfrastructureΠεριβάλλον δοκιμών
A facility or environment for trying out a technology under defined, realistic conditions. It can combine equipment, simulation and controlled data. Its main purpose is technical testing, which sets it apart from a regulatory sandbox.
ExampleA delivery robot is tested around obstacles and people in a controlled space before a trial in a real workplace.
AI basicsΤολμηρό ερευνητικό εγχείρημα
An ambitious research effort with a substantial chance of failure and a potentially large payoff. Calling such research “high-risk” refers to that chance of failure and is unrelated to the risk categories of the EU AI Act.
ExampleA university team tries a new way for AI to learn from very little data, with set milestones for checking whether the approach works.
AI basicsΕγγραμματισμός στην ΤΝ
The knowledge and judgement needed to use AI sensibly: recognising its limits, checking what it produces and understanding how it affects people.
ExampleA secondary-school teacher checks an AI-generated summary against the original source and keeps pupils’ personal details out of the tool.
AI basicsΡομποτική
Machines that sense or act in the physical world. Some follow fixed instructions, while others use AI to interpret their surroundings and decide what to do.
ExampleA warehouse robot may use AI to find its way between the shelves, while its lifting mechanism follows programmed rules.
Regulation and governanceΚυβερνοασφάλεια
Protecting digital systems, services and information from disruption, unauthorised access and misuse. AI can help defenders and can also introduce new weaknesses, so basic security controls and the ability to recover from an incident still matter.
ExampleA family-run hotel in Ayia Napa limits who can open guest records, installs updates promptly, watches for unusual activity and tests its backups, so it can restore its booking system after an incident.
AI basicsΤεχνητή νοημοσύνη
Computer systems that use the information they receive to produce predictions, content, recommendations or decisions. Each system is designed for particular tasks.
ExampleAn email service marks an incoming message as likely spam and moves it out of the inbox.
AI basicsΠαραγωγική ΤΝ
AI that creates new content, such as text, images, audio or computer code, from patterns it learned during training.
ExampleThe owner of a guesthouse in Paphos uses an AI tool to draft a welcome email for guests and reads it through before sending.
AI basicsΜοντέλο ΤΝ
The part of an AI system that has learned from data and turns inputs into outputs. A foundation model is trained broadly so that it can be adapted to many tasks. The indicator of AI models developed in Cyprus is about building models, which is a separate capability from using a model made elsewhere.
ExampleAn online shop in Nicosia uses a model built abroad to answer customer questions about its own products.
AI basicsΜεγάλο γλωσσικό μοντέλο
An AI model trained on large amounts of written language so that it can generate and work with text. Chatbots are a common use.
ExampleA civil servant asks a chatbot to summarise a long public consultation document, then checks the summary against the original.
AI basicsΨευδής παραγωγή περιεχομένου
An AI answer that sounds plausible but is false or has no support in any source. It can include invented facts, quotations or references.
ExampleAsked for studies on tourism in Cyprus, a chatbot gives a convincing title and web link for a report that does not exist.
InfrastructureΥπολογιστική ισχύς
The computing resources used to train and run AI: processors, memory and the time they spend working. Who can use them, the type of hardware and the cost all matter, as well as the number of machines.
ExampleA university research team applies for time on a shared supercomputer instead of buying its own hardware.
InfrastructureGPU / επιταχυντής
A processor that carries out many calculations at the same time, which suits it to training and running AI. A count of GPUs shows how much equipment there is, and different models can differ greatly in performance.
ExampleTwo computing centres with the same number of GPUs take different amounts of time to finish the same task, because one has newer models.
InfrastructureΕργοστάσιο ΤΝ
A European initiative that brings together supercomputing resources and expert support for developing AI. In this name, “factory” means a hub offering computing and related services.
ExampleA start-up uses time on a hub’s supercomputer, with advice from its specialists, to develop an AI application.
InfrastructureΚεραία Εργοστασίου ΤΝ
A local support hub that connects people and organisations to an AI Factory elsewhere. The supercomputer remains at the AI Factory, and the Antenna gives local users a route to it.
ExampleLocal specialists help a research team prepare an application for time on a supercomputer in another country.
Statistics and measurementΥιοθέτηση ΤΝ
How widely AI is actually used. The main adoption figure on this site covers enterprises with ten or more people employed, in the sectors that Eurostat’s survey includes.
ExampleA self-employed architect in Larnaca uses an AI assistant every day but falls outside the figure, which starts at ten people employed.
Statistics and measurementΒασικές ψηφιακές δεξιότητες
The everyday ability to use digital tools for finding information, communicating, creating content, staying safe and solving problems. The measure is broader than the ability to use AI.
ExampleA retired teacher in Paralimni checks that health advice comes from a reliable website, then shares a document with her daughter so that only she can open it.
Statistics and measurementΠοσοστιαία μονάδα
The simple difference between two percentages. It is the usual way to describe a gap between two shares, such as the share of enterprises using AI in Cyprus and in the EU.
ExampleIf a share rises from 10% to 15%, it has grown by 5 percentage points, or by 50% in relative terms.
Statistics and measurementΣυνολική βαθμολογία
A score from 0 to 100 that summarises selected indicators. Each figure is first scored against its published reference band, and the metric scores are then combined using stated weights. The result depends on which indicators are included and how they are weighted.
ExampleAn illustrative score of 40 out of 100 shows where the figures sit within their reference bands, taken together. It does not mean that a country is “40% ready for AI”.
Statistics and measurementΒαθμίδα ποιότητας
A label showing how a figure was produced. Tier A (measured) is published by the responsible authority, tier B (modelled) is derived using a documented method, and tier C (estimate) is an approximation. How recently a figure was checked is shown separately, by its review date.
ExampleA tier C estimate checked last week is still tier C. Its recent review date shows that the figure is up to date, and its tier shows that it is an estimate.
Statistics and measurementΤελευταία ελεγμένη τιμή
The most recent figure we checked against its source and published. It is shown when an automatic update from the source is not available, and its review date tells you how current it is.
ExampleWhile the publisher’s data service is down, a chart keeps showing the survey results we last checked, with the date of that check.