Weekly Review 17 December 2021

Some interesting links that I Tweeted about this week:

1) The computational demands of AI continue to outstrip processing power: https://spectrum.ieee.org/ai-training-mlperf

2) Using AI to play Pokemon: https://www.theverge.com/2021/12/2/22811849/pokemon-competitive-ai-nintendo-game-freak Is this really AI, or just a brute-forcing of options?

3)A list of popular libraries for Machine Learning: https://www.datasciencecentral.com/profiles/blogs/a-hand-picked-list-of-top-python-frameworks-and-libraries-for I've been using sklearn to teach my #AI class.

4) Managing long COVID with AI https://dataconomy.com/2021/12/how-artificial-intelligence-long-covid/

5) How to have trust in AI, when most AI is biased: https://www.techrepublic.com/article/how-much-can-we-trust-ai-how-to-build-confidence-before-large-scale-deployment/

6) Best practices for implementing Natural Language Processing. Good principles for most #AI applications, actually: https://www.informationweek.com/big-data/6-best-practices-for-nlp-implementation

7) Tips and code for time-series forecasting: https://www.kdnuggets.com/2021/12/avoid-mistakes-time-series-forecasting.html

8) Bad AI models are bad for business: https://www.datanami.com/2021/11/29/youve-been-warned-bad-data-models-are-capable-of-destroying-companies/

9) More moves to increase the accountability of using AI https://arstechnica.com/tech-policy/2021/12/the-movement-to-hold-ai-accountable-gains-more-steam/

10) Protein folding prediction is a hard problem. If AlphaFold has solved it, that's a major breakthrough: https://www.technologyreview.com/2021/07/22/1029973/deepmind-alphafold-protein-folding-biology-disease-drugs-proteome/

11) Most of the research outputs from universities a built on work done by PhD students. With the part-time job losses from the pandemic, either PhD stipends increase, or research outputs are going to suffer: https://www.odt.co.nz/opinion/phd-stipend-must-keep-living-costs

12) Biased data leads to biased models. When we used biased data sets, we get biases results. Why is this hard for people to grasp? https://www.technologyreview.com/2021/04/01/1021619/ai-data-errors-warp-machine-learning-progress/

13) Machine Learning engineers are now more important that data scientists: https://www.kdnuggets.com/2021/11/why-machine-learning-engineers-are-replacing-data-scientists.html

14) A German case, but with more and more people expected to keep working from home after the pandemic, such cases will probably become more common: https://www.theregister.com/2021/12/10/bed_to_desk_workplace_accident/

15) A look at the first computer-composed music: https://www.theguardian.com/music/2021/dec/07/he-touched-a-nerve-how-the-first-piece-of-ai-music-was-born-in-1956