AI wrap Friday June 16th
Thanks to everyone who shared the links in this post with me this week while I am on holiday, must say I have had less time reading than anticipated - more time swimming and sunbathing.
This is also publishing a tad earlier due to timezones. Hope you enjoy this wrap of AI in the news, Friday June 16th 2023. Kia pai tō rā, Vic
What are the UK government up to AI wise?
Back in March the UK Government published a white paper on their approach to AI, continuing their theme of innovation, it’s a great foundation document. I was surprised their consultation only received feedback from 130 respondents given the size of their industry. They lay out their plans in 6 month windows, which includes developing a roadmap early on.
Roll forward to this week. Prime Minister Rishi Sunak announced a deal with Google DeepMind, OpenAI and Anthropic to open up their models to the UK government for “research and safety purposes”, “to help build better evaluations and help us better understand the opportunities and risks of these systems”.
Love that there is a PM embracing AI as transformational for his country especially in healthcare and education spaces while ensuring there are some protections in place - hopefully this is just a starting point, and that UK government will take this seriously and not just blindly accept AI enhancements from any and all tech companies without the same protections in place.
All of that said, we can learn a lot from what the UK are doing here.
AI and Ethics
In a week when Margrethe Vestager, the European Commission for Competition said “Discrimination is a more pressing concern from advancing artificial intelligence than human extinction” these two initiatives from the World Economic Forum and UNESCO are very timely.
UNESCO launches Women4Ethical AI expert platform to advance gender equality.
Very cool initiative we can all embrace. “Today, globally, women and girls are 25% less likely than men to know how to leverage digital technology for basic purposes, 4 times less likely to know how to programme computers and 13 times less likely to file for an ICT patent.” With other stats like 6% of professional software developers are women UNESCO is taking a proactive approach to equality for women with this initiative.
World Economic Forum on AI and access to education.
The WEF has turned it’s sights on education and training systems which must “keep pace with the AI transformation to ensure future-ready and inclusive curricula”. The article provides a great framework and goes on to state publicly accessible training and education must be on the national AI agenda citing examples like Egypt who is teaching students about AI in classroom already, Germany’s apprenticeship model providing hands on experience and selected other national programmes. Again, Aotearoa NZ could - and should - learn a lot from this.
Also worth reading is this - the looming algorithmic divide.
AI and the environment
Is Generative AI bad for the environment?????? A nice blog from Boston University Associate Professor of Computer Science, Kate Saenko. She explains the concepts really nicely and provides insight into the environmental cost of the models we know today, just building GPT-3 (not running it) cost the “equivalent of running 123 gasoline-powered passenger vehicles for one” year for instance. She concludes it will be down to us, the consumers, to choose to use greener technology and encourage the tech giants to publish their carbon footprint stats.
AI and Energy. This scientific journal article on “Applications of Artificial Intelligence Algorithms in the Energy Sector” contains great definitions for common terms as well as laying out the case for the role of AI in Smart Grids and intelligent energy management systems.
ChatGPT’s jokes
ChatGPT repeats the same 25 jokes a lot it seems. Some people just have more time on their hands, German researchers found "Over 90% of 1,008 generated jokes were the same 25 jokes." Why they were researching this who knows but it is amusing in itself. And ChatGPT’s favourite joke?
Q: Why did the scarecrow win an award?
A: Because he was outstanding in his field.
Who pays what?
Reddit haven’t covered themselves in glory recently but I do sympathise with their reaction to OpenAI’s process of hoovering up their platform content to train ChatGPT, declaring they will charge fees for anyone who wants to take this approach. One platform Apollo suggests “Reddit's pricing is punitive and aimed particularly at AI companies that need large volumes of training data.” And it will cost them a massive $20M per annum to keep running as is - again my sympathy lies with Reddit on this one, if you are building a product or making money using someone else’s platform then you should pay for the use, surely? This is all wrapped up in their new 3rd party access data charging model which has shutdown loads of subreddits this week.
This article sums it all up nicely. In the words of Taylor Swift, "Karma's gonna track you down."
In short
- Want to improve your online dating profile and success rate? Meet Millie claims “Get 100% response rate on dating app”. Wow, but no I have not tried it.
- Something to keep an eye on is this - “AI hype could end up like the dot-com bubble, veteran investor warns”
- Codeprompt is a code generator for Javascript, Typescript and React
And finally, Sir Paul McCartney is creating one more song by The Beetles with the help of AI technology which will include John Lennon’s voice. There’s not much info out there on this yet, here is a link to a CNN interview.