New study pinpoints the weaknesses in AI
11.1.2024 09:52:51 CET | Københavns Universitet | Pressemeddelelse
ChatGPT and other solutions built on Machine Learning are surging. But even the most successful algorithms have limitations. As the first in the world researchers from University of Copenhagen has proven mathematically that apart from simple problems it is not possible to create algorithms for AI that will always be stable. The study may lead to guidelines on how to better test algorithms and reminds us that machines do not have human intelligence after all.

Machines interpret medical scanning images more accurately than doctors, they translate foreign languages, and may soon be able to drive cars more safely than humans. However, even the best algorithms do have weaknesses. A research team at Department of Computer Science, University of Copenhagen, tries to reveal them.
Take an automated vehicle reading a road sign as an example. If someone has placed a sticker on the sign, this will not distract a human driver. But a machine may easily be put off because the sign is now different from the ones it was trained on.
“We would like algorithms to be stable in the sense, that if the input is changed slightly the output will remain almost the same. Real life involves all kinds of noise which humans are used to ignore, while machines can get confused,” says Professor Amir Yehudayoff, heading the group.
A language for discussing weaknesses
As the first in the world, the group together with researchers from other countries has proven mathematically that apart from simple problems it is not possible to create algorithms for Machine Learning that will always be stable. The scientific article describing the result was approved for publication at one of the leading international conferences on theoretical computer science, Foundations of Computer Science (FOCS).
“I would like to note that we have not worked directly on automated car applications. Still, this seems like a problem too complex for algorithms to always be stable,” says Amir Yehudayoff, adding that this does not necessarily imply major consequences in relation to development of automated cars:
“If the algorithm only errs under a few very rare circumstances this may well be acceptable. But if it does so under a large collection of circumstances, it is bad news.”
The scientific article cannot be applied by industry for identifying bugs in its algorithms. This wasn’t the intension, the professor explains:
“We are developing a language for discussing the weaknesses in Machine Learning algorithms. This may lead to development of guidelines that describe how algorithms should be tested. And in the long run this may again lead to development of better and more stable algorithms.”
From intuition to mathematics
A possible application could be for testing algorithms for protection of digital privacy.
”Some company might claim to have developed an absolutely secure solution for privacy protection. Firstly, our methodology might help to establish that the solution cannot be absolutely secure. Secondly, it will be able to pinpoint points of weakness,” says Amir Yehudayoff.
First and foremost, though, the scientific article contributes to theory. Especially the mathematical content is groundbreaking, he adds:
”We understand intuitively, that a stable algorithm should work almost as well as before when exposed to a small amount of input noise. Just like the road sign with a sticker on it. But as theoretical computer scientists we need a firm definition. We must be able to describe the problem in the language of mathematics. Exactly how much noise must the algorithm be able to withstand, and how close to the original output should the output be if we are to accept the algorithm to be stable? This is what we have suggested an answer to.”
Important to keep limitations in mind
The scientific article has received large interest from colleagues in the theoretical computer science world, but not from the tech industry. Not yet at least.
”You should always expect some delay between a new theoretical development and interest from people working in applications,” says Amir Yehudayoff while adding smilingly:
”And some theoretical developments will remain unnoticed forever.”
However, he does not see that happening in this case:
”Machine Learning continues to progress rapidly, and it is important to remember that even solutions which are very successful in the real world still do have limitations. The machines may sometimes seem to be able to think but after all they do not possess human intelligence. This is important to keep in mind.”
The scientific article ”Replicability and Stability in Learning”, was published at the Foundations of Computer Science (FOCS) conference 2023.
Kontakter
Amir Yehudayoff
Professor
Department of Computer Science (DIKU)
University of Copenhagen
amye@di.ku.dk
+45 93 89 53 88
Michael Skov Jensen
Journalist and team coordinator
The Faculty of Science
University of Copenhagen
Mobile: + 45 93 56 58 97
msj@science.ku.dk
Følg pressemeddelelser fra Københavns Universitet
Skriv dig op her, og modtag pressemeddelelser på e-mail. Indtast din e-mail, klik på abonner, og følg instruktionerne i den udsendte e-mail.
Flere pressemeddelelser fra Københavns Universitet
Nyt fund omskriver historien om en af menneskets vigtigste forfædre19.11.2026 07:00:00 CET | Pressemeddelelse
Fossile tænder fra to huler i Kina viser, at menneskets forfader, Homo erectus, levede på det asiatiske fastland langt senere end hidtil troet. Forskerne fandt desuden en tand fra en anden menneskeart, der levede side om side med Homo erectus og det udfordrer vores forståelse af menneskets evolution.
Hvor mange job vil AI koste? Forskere advarer mod forhastede konklusioner18.11.2026 06:05:00 CET | Pressemeddelelse
Nyt CAISA-brief advarer mod at forveksle ’AI-eksponering’ med ’jobtab’ eller ’sparede årsværk’. Ifølge forskerne bag briefet er vi i den offentlige debat tilbøjelige til at overse de komplekse mekanismer, der afgør, hvordan kunstig intelligens faktisk påvirker beskæftigelsen.
Breakthrough: Danish researchers find efficient solution to famous 174-year-old mathematical headache17.11.2026 11:50:37 CET | Press release
Can any map be coloured using just four colours? That is the question behind one of the most famous problems of mathematics, dating back to 1852. Since 1976, we have known that the answer is ‘yes’ – but how do you actually find the coloring? Now, two professors from the University of Copenhagen and DTU, together with international colleagues, have found the optimal algorithm after a decade-long ‘obsession’. The answer provides new fundamental mathematical insights.
Gennembrud: Danske forskere har fundet en effektiv løsning på 174 år gammel matematisk hovedpine17.11.2026 10:58:34 CET | Pressemeddelelse
Kan ethvert landkort farvelægges med blot fire farver? Det handler et af matematikkens mest berømte problemer om, som stammer fra 1852. Vi har siden 1976 vidst at svaret er 'ja', men hvordan gør man det? Nu har to professorer fra Københavns Universitet og DTU sammen med udenlandske kolleger fundet en optimal algoritme, og det efter 10 års ’besættelse’. Svaret giver nye grundlæggende matematiske indsigter.
High water, high rents: Poorest residents pay the most for climate adaptation12.11.2026 07:17:00 CET | Press release
As cities worldwide adapt to extreme weather, climate adaptation must prevent people from wading through flooded streets in the future. But the costs are not equally shared. A study from the University of Copenhagen shows that climate adaptation policies may even make social inequality worse. In Copenhagen, stormwater costs fall disproportionately on not-for-profit housing residents, as these neighborhoods are often located near climate adaptation projects.
I vores nyhedsrum kan du læse alle vores pressemeddelelser, tilgå materiale i form af billeder og dokumenter samt finde vores kontaktoplysninger.
Besøg vores nyhedsrum