Computing exams, AI, water and datacentres – Pc Weekly Downtime Upload podcast

In this episode of the Pc Weekly Downtime Add podcast, Clare McDonald, Brian McKenna and Caroline Donnelly talk about the 2021 A-degree and GCSE computing results, what algorithms are and usually are not good for, and the water consumption habits of datacentres, and their environmental influence.

Clare opens up the episode with an account of the newest A-degree and GCSE leads to England, Wales and Northern Ireland – a bumper crop of examination outcomes that are normally a bit more spaced out.

This yr’s results have been algorithm-free, in contrast to, initially, the 2020 outcomes. But not freed from their standard class bias: personal faculty college students did even higher than traditional.

Clare notes the nerve-racking Covid context for this yr’s college students. Despite that, there was a better-achieved set of A-degree computing outcomes. There was a rise in the number of students taking A-degree computing: 13,829 college students in the UK took computing at A-degree, a rise from 12,428 entrants the previous yr.

There was a yr-on-yr improve within the variety of women taking A-degree computing but in addition, worryingly, a drop in the numbers taking GCSE. However, women are doing better. For women, 25.7% achieved an A* outcome, a rise from 17.eight% final yr; whereas solely 18.9% of boys achieved an A* degree grade, an increase from thirteen.1% final yr. For the primary time, women also outperformed boys in mathematics in A-ranges and GCSEs.

Algos – what are they good for?

The episode then strikes on to a related dialogue about what algorithms and synthetic intelligence (AI) are good for, if they don’t seem to be good for exams.

Brian touches on the A-degree and Scottish Highers debacle of 2021, when algorithms received themselves a nasty identify. This was mentioned on the podcast virtually precisely a yr ago.

The entire fiasco was talked about in a newer BCS report, Priorities for the national AI technique, written by Invoice Mitchell, the BCS’s director of policy.

The report says the UK can take a world lead in AI ethics if it cultivates a extra numerous workforce, together with individuals from non-STEM backgrounds. It refers again to the Ofqual algorithm that was used to estimate GCSE and A-degree grades in 2020. In the BCS writer’s phrases, this led to a “widespread public distrust in algorithms making high stakes selections about individuals”.

The report registers that “public belief in AI and algorithmic techniques normally has been significantly eroded by occasions in the course of the pandemic”, as shown in two national surveys by YouGov commissioned in 2020 by BCS. These discovered that:

  • Over half (53%) of UK adults haven’t any faith in any organisation to make use of algorithms when making judgements about them, in issues starting from schooling to welfare selections.
  • sixty three% of UK adults disagree with the statement “College students graduating with a computer science college diploma are certified to write down software that makes life selections about individuals”.

The BCS report itself will feed into the government’s AI technique, which shall be revealed later this yr.

On the podcast, Brian says the headline in it for him was the concept for the public to belief AI techniques, we’d like a broader set of people making them within the first place. Not that it is all bleak for algorithms. Where would we be with out the recommendation algorithms of the streaming providers we have now trusted a lot over the pandemic?

Brian then poses the question: are there areas of human life that should just be free of algorithms?

Caroline cites one space that must be off-limits to algorithms and knowledge analytics: Queen of Pop Beyoncé’s profession, as illustrated in an interview in Harper’s Bazaar. Beyoncé’s refusal of knowledge analytics-based mostly recommendation relating to the 2008 album I am…Sasha Fierce is an exemplar of preserving with “the human feeling and spirit and emotion in my choice-making”, she says.

The group talk about another areas the place AI could possibly be inappropriate, reminiscent of recruiting, but in addition might be useful. We might, despite the huge hype, be in the early days of how AI will rework human lives. That is, if people survive local weather disaster.

Water and datacentres: a matter beneath-discussed in local weather change debates

Within the third section of the episode, Caroline discloses some work progress on the subject of the water consumption habits of datacentres, and what those might mean for the setting in the context of the rising climate disaster.

She reveals how her consideration was initially drawn to this theme by a remark at a trade show in 2016 – that it isn’t recognized how a lot water is used by datacentre operators in their cooling techniques.

Datacentres have approach beneath scrutiny for their environmental influence and sustainability efforts along three dimensions, explains Caroline: how a lot energy datacentres use, how much of that energy is renewable, and the way massive the carbon footprint of datacentres is. Progress has been made on these three fronts.

There’s a extensively used business metric for measuring the power consumption of datacentres, PUE – the power usage effectiveness rating.

And whereas datacentre operators are advantageous about disclosing their PUEs, they’re oddly silent about another measure – their water utilization effectiveness. There is a metric, revealed in 2010, that the datacentre business might use for the water utilization effectivity of their datacentres – WUE (water utilization effectiveness). But it’s both not being used, or the scores are usually not being revealed. Fb is a partial exception, however that’s of little second to enterprise IT consumers.

Caroline cites an Uptime Institute survey from 2020 that stated half of all datacentre operators don’t measure how much water they use. Might this be because operators are utilizing plenty of water to deflate their PUE scores?

Water is a valuable resource, particularly so in regions which are increasingly subject to drought, corresponding to California, or which endure a basic stress on the supply of consuming water. Those embrace nations similar to Spain and Singapore. And that roster of water-burdened areas might get greater in years to advance – datacentres are lengthy-time period features of the landscape.

Recent air cooling might be part of the solution to the over-use of water. Another could possibly be the datacentre-on-a-barge strategy being pioneered by Nautilus Knowledge.

It’s early days for this matter, says Caroline: “Water is something that datacentre operators aren’t talking about, however with local weather change, and the threat of that turning into even more apparent and real than it beforehand was, that should change. And there must be extra strain on operators to be more transparent about how what they do affects water provides.”

Podcast music courtesy of Joseph McDade

Translate »