In a\u00a0nondescript grey\u00a0building,\u00a0nestled\u00a0between the\u00a0rolling\u00a0hills and green trees of\u00a0Penicuik on the outskirts\u00a0of Edinburgh,\u00a0sits\u00a0one of the most important\u00a0computing\u00a0facilities in the UK.\u00a0<\/p>\n
It\u2019s\u00a0an unassuming\u00a0location for such an influential\u00a0place, but the Advanced Computing Facility (ACF),\u00a0operated\u00a0by the EPCC\u00a0(formerly the Edinburgh Parallel Computing Centre)\u00a0at the University of Edinburgh, is the nerve centre of the\u00a0UK\u2019s\u00a0past, present and future\u00a0in supercomputing<\/a>.<\/p>\n Housed at the site is\u00a0ARCHER2,\u00a0one of\u00a0the UK\u2019s most advanced\u00a0supercomputers,\u00a0capable of solving complex equations\u00a0in a fraction of a second. Plans are in motion to build\u00a0its\u00a0replacement, a\u00a0next-generation\u00a0supercomputer<\/a>\u00a0that the UK Government\u00a0is\u00a0funding to the tune of \u00a3750m.\u00a0<\/p>\n Heading up this operation\u00a0is\u00a0Professor Mark Parsons\u00a0who,\u00a0as director of the EPCC, is\u00a0responsible\u00a0for\u00a0implementing and running\u00a0a project that could be crucial in ensuring the UK\u00a0doesn\u2019t\u00a0fall behind in the computing arms race. \u00a0<\/p>\n \u201cI’m\u00a0really proud\u00a0of the fact that EPCC has survived and flourished since we were set up in 1990,\u201d says\u00a0Parsons. \u201cGetting that\u00a0\u00a3750m\u00a0of\u00a0funding is\u00a0really important\u00a0because we were going to end up as a country without any national supercomputing resource and that would just be wrong.\u00a0There’d\u00a0be no\u00a0European country like it.\u201d\u00a0<\/p>\n Inside\u00a0the\u00a0walls of the ACF, rows upon rows of black\u00a0cupboards house\u00a0thousands of computer chips, all cooled by a complex system of pipes\u00a0that circulate\u00a0cold water\u00a0into the computer and hot water back out.\u00a0The pipes are used\u00a0to keep\u00a0ARCHER2\u00a0as cool and\u00a0efficient as possible\u00a0while\u00a0it works away\u00a0solving any number of complex problems that a standard system would\u00a0flounder at the metaphorical sight of.\u00a0<\/p>\n The\u00a0future\u00a0system at the\u00a0ACF, which is predicted to come online in 2027 after next-generation\u00a0chips\u00a0from Taiwan are\u00a0purchased, is predicted to be\u00a050 times faster than ARCHER2. This will open the door for a new level\u00a0of computational processing that allows for even larger and more difficult problems to be solved. \u00a0<\/p>\n \u201cI’ve got a big ten-year\u00a0project with Rolls-Royce that we’re halfway through, where we’re starting to model whole gas-type turbine engines,\u201d says Parsons.\u00a0\u201cWe\u00a0can’t\u00a0do that on ARCHER2.\u00a0It\u00a0doesn\u2019t\u00a0have\u00a0the\u00a0memory\u00a0and\u00a0it’s\u00a0not fast enough.\u00a0But\u00a0we’ll\u00a0be able to do that on the next system.\u201d\u00a0<\/p>\n Taiwan dominates the production of semiconductors, the\u00a0single most crucial\u00a0component\u00a0needed to make a supercomputer.\u00a0These semiconductors are used to make the silicon chips that\u00a0act as the \u201cbrains\u201d for modern computers.\u00a0Companies based\u00a0in Taiwan\u00a0produce\u00a0at least\u00a090 per cent of\u00a0the world\u2019s most advanced\u00a0semiconductors and generate\u00a0over $125bn in annual revenue\u00a0from the sale of them to countries around the world. \u00a0<\/p>\n \u201cI don’t think we as a country have, at the moment, the infrastructure to produce silicon chips,\u201d says Parsons. \u201cBut we will\u00a0[in the future]\u00a0have\u00a0enough\u00a0significant infrastructure that can deliver the sort of projects and new activities\u00a0that our public sector,\u00a0industry\u00a0and academia need to do.\u00a0So,\u00a0our\u00a0computer is just part\u00a0of that path.\u201d\u00a0<\/p>\n The\u00a0new supercomputer\u00a0at the ACF\u00a0will be the centrepiece of a UK-wide scheme, announced by the UK Government<\/a> and costing \u00a31bn, that should increase the country\u2019s computing capacity 20-fold in the next five years. In comparison, China has announced a plan to invest one trillion yuan (\u00a3105bn)\u00a0in\u00a0a government-backed fund focused on increasing the country’s global competitiveness in\u00a0technology\u00a0such as\u00a0quantum computing and the ever-evolving\u00a0world of artificial intelligence\u00a0(AI).\u00a0<\/a><\/p>\n The\u00a0emergence of AI as a significant factor\u00a0in our lives\u00a0has been powered by the\u00a0supercomputers\u00a0that facilities like the ACF\u00a0host. These systems have been used to\u00a0train large language models\u00a0(LLMs)\u00a0like ChatGPT,\u00a0a\u00a0system\u00a0used\u00a0daily\u00a0by\u00a0over 800 million\u00a0people\u00a0to request answers,\u00a0write\u00a0poems\u00a0and generate silly little images of cats in party hats.\u00a0Despite this, Parsons thinks that many of the big investments into computing taking\u00a0place\u00a0around\u00a0the world\u00a0aren\u2019t\u00a0focused on the training of AI\u00a0for future usage\u00a0but more on the running of current\u00a0AI chatbots\u00a0to answer queries\u00a0on known datasets.\u00a0<\/p>\n \u201cIt’s\u00a0fantastic\u00a0we’ve\u00a0got large language models, they’re\u00a0just fascinating things to interact with,\u201d says Parsons. \u201cBut\u00a0actually, the use of AI is much, much broader than that and many of the benefits of it will not be through a large language model.\u201d\u00a0<\/p>\n The potential for AI to increase productivity, cut costs for businesses and create new\u00a0opportunities for growth in the economy\u00a0is\u00a0balanced by\u00a0real\u00a0fears over its implementation at the cost of human jobs and society. In polling published by the UK Government, 33 per cent of respondents\u00a0said that they expected the future impact of AI on society to be a negative one.\u00a0\u00a0<\/p>\n \u201cIt’s\u00a0clear that when\u00a0we’ve\u00a0had moments in time where there have been\u00a0big changes\u00a0in technology,\u00a0the\u00a0Industrial Revolution is the biggest example of that,\u00a0where\u00a0people’s jobs have changed,\u201d says Parsons.\u00a0\u201cBut if you look back through history, the number of jobs\u00a0has\u00a0not got\u00a0less\u00a0generally, people\u00a0have just changed.\u00a0So,\u00a0I think for many people, AI is just going to help them with what they do\u00a0currently.\u201d\u00a0<\/p>\n This\u00a0doesn\u2019t\u00a0mean the technology\u00a0is\u00a0without flaws.\u00a0Studies show that AI systems\u00a0like ChatGPT\u00a0can \u201challucinate\u201d answers to questions that\u00a0it does not know, providing users with incorrect information<\/a>.\u00a0AI systems\u00a0can also be influenced by the\u00a0data\u00a0used\u00a0to\u00a0train\u00a0them\u00a0if\u00a0that data is incorrect or lacking context and depth.\u00a0<\/p>\n \u201cMy\u00a0father\u00a0is\u00a0in his late 80s now and was a famous geologist,\u201d says\u00a0Parsons.\u00a0\u201cI sat with\u00a0him\u00a0and we were playing with ChatGPT,\u00a0asking it really complicated geology questions, which he was amazed\u00a0that\u00a0it was getting it right.\u00a0But then we asked about one of his areas of research that he really focused on in the 1970s and it knew nothing about it at all. And it’s\u00a0because those papers have never been digitized,\u00a0they’re\u00a0just available on paper.\u201d\u00a0<\/p>\n Without the full context,\u00a0Parsons\u00a0worries that we could be entering a world of \u201cbland internet culture\u201d that lacks the local\u00a0knowledge available in\u00a0undigitised\u00a0resources. A world of information\u00a0where answers are given on a global scale,\u00a0without\u00a0taking into consideration the in-depth, region-specific context that\u00a0is necessary to provide useful\u00a0and\u00a0accurate\u00a0information\u00a0to users.\u00a0<\/p>\n \u201cThere’s\u00a0a pre-internet dark age where none of that information is in these models,\u201d says Parsons.\u00a0\u201cSome really old\u00a0information is\u00a0because there are electronic versions of the works of Shakespeare and Jane Austen.\u00a0But there\u00a0are\u00a0big\u00a0bits of our\u00a0knowledge that\u00a0have\u00a0never been\u00a0digitised\u00a0and I think that’s\u00a0a danger with where some of the models are going.\u201d\u00a0<\/p>\n Holyrood Newsletters<\/p>\n