Dear Authors AI Dilemma
I began co-editing intergenerational futures to 1984’s start of the 2025 report genre https://www.mattridley.co.uk/blog/death-of-a-great-optimist/
In my father’s last years diffusing his archives became full time work with eg Glasgow University Adam Smith scholars taking this up and designing 2 new journals: Social Business and New Economics. Then Japan Ambassador started 16 trips of mine to Bangladesh discussing context of Neumann’s, dad’s and Fazle Abed’s life work empowering women. Culturally, my learning curve has involved mediating East-West Exchanges with 60 trips to Asia which started when I was contracted for 8 years on Unilever Muslim women’s market needs beginning with Indonesia. As typical with Diaspora Scots families, full circles of life's purpose came for dad by surviving being teenage navigator Allied Bomber Command World War 2 but by 1962 celebrating Deming’s reinvigoration of Japan quality engineering www.economistjapan.com as start of Asian Peoples Rising
Diversity is simpler to vision bottom-up. At Fazle Abed's 80th birthday party spring 2016, partners saw SDG goals 1-5 (part of 6) as linked fields which Asia’s 45-year female intel miracle networked, where necessary without electricity, but also blueprinting leapfrog models partnered by solar and mobile. But, Goals 16 down require infrastructure cooperation. Abed suggested transforming university partners. Where mapping sdg factors across LLM - goal 17 is questionable. in data depth and millennial trust. New 17 could be Human AIGood and interact whole UN2 framework. How to mediate exchanges between 2 frameworks fits UN Global Futures Year if top level Guterres AI panel is asked how to LLM the 2 systems together as one. There is valuation support from UN egov ranking tables and other multilateral convergences. Digital identity eg India’s experience another key especially if multilateral AI unites https://unsceb.org/board-members
As a statistician, I am experienced at deep data mapping - ie seeing if something is being left out by a computational model compared with blueprint that eg bottom 2 billion peoples used when they had no advanced analytics.. My book Brand Chartering of organisiation-wide learning is the benchmark for integrating disciplines and stakeholder flows. Before Economist Intelligence Unit publication, it was proof-tested over 5 years in the world’s largest advertising agencies valuing integrity of corporate brands. Whilst I have not personally trained a computer in the imagenet way, I have enough analytic experience to see if a huge amount of computing time is being given the right data to start with. My understanding is your high-level group including Fei-Fei Li alumni are over 18 years valuing computer vision as www freedom. For over 30 years now I have worked in team projects; agreeing who is best at doing what is also about not wanting anyone to claim to be cutting edge at everything
ee intel tests at bard.solar . Who'll win race as LLM of 1000 languages (Pichai)? Bard is frank about all LLM races as gateways to web3 having good & bad players. 23-24 is critical with Global Futures Forum briefing UN started this Spring. If we can get enough youth envoy time testing cooperation LLM,, we can wiin. Ecosystem of Stanford and Mountain View unique in West. From that space, Bard says UN needs LLM plugin its education, training & outreach last mile practices. NB November’s women empowerment team launching Li’s" Worlds I See" Li, MGates, Condoleezza Rice, JDowds, PChan. Their LLM focus links place governance & girls & health & edtech. And they see whats on inside big 6 digital companies ieg offering smaller or older models as open while running big private tests Back in 1984, our 2025 Report made case that personalized learning was key to millennials being the sdg generation. UNGA on edu unfit for purpose happened before LLM’s all change opportunity. Time to revision.
Stanford HAI publishes world class resources eg foundations models, annual AI index report. Assume Fei-Fei Li as great a mathematician as Von Neumann - her responsibility for vision coding spans all manifestations of which LLM is just latest. I was in the room for 7 hours Spring 2023 as Fei-Fei Li completed year 1 of America's https://www.ai.gov/nairrtf/ . Coding vision data since 2006 was so much work it had to be both open & orchestrated by a few leaders eg by 2010 Fei-Fei Li, Ng and Hassabis. But now they have a fight as big companies face loss of their dominance of web2. Like my #1 hero Fazle Abed, she doesn’t want high visibility for what mass of millennials need to spend their time and data on. She does not court celebrity but, In my opinion, a lot of mutual respect needs to grow across UN and Fei-Fei Li's core HAI nets. Dynamic similar to Von Neumann 1951 but hopefully this time, a maths genius with a normal life span ahead to keep inspiring humans on our augmentation orbits.
Dad (Norman Macrae) was biographer of Von Neumann with tech research by me. Recently daughter Marina inspired friends to pilot AI Games: "Architect Intelligence" www.AIGames.solar . Who'd John celebrate as advancing humanity with computing machines? We see John celebrating Dr Fei-Fei Li as 21st C maths wizard. Cooperative coding of machine vision around objects that humans see and work with is different in consequence to binary program languages. It bridges Von Neumann's “The Computer and The Brain”. Today AI4All shows youth can begin AI lifestyle with Dance or Arts needing no maths and mobilised by apps. Thus, LLMs with plugins are the gateway to web3. Unlike web2's race between big corporates, this is a game curious youth and teachers and community culture can play, as can UN AI edu. See Nov 2023 book: "Worlds I See" in Melinda Gates library “Moment of Lift” for SDG1 poverty alleviation by empowering women-led communities, the way ahead, eg Fazle Abed partners catalogue abedmooc.com
Please read statement in conjunction with answers to 8 questions framing this special project.
I recommend clarifying human intelligence ethics before personifying ethical risks of AI.
Engineer David Weir, who headed Bradford Business School in 1995, appointed me as senior visiting fellow to form a second opinion from a media viewpoint on set of disaster cases from an engineering perspective. In most cases, the PR and lawyer professions had covered up some real causes that would either have damaged leadership reputations or caused far larger payouts by insurance companies. This prepared me for informal analysis of 2000s disasters: 9/11, dotcom audit failures due to Andersen and Chicago short-term mindsets, 7/7 and subprime. Societal learnings locally and globally were not continuously mapped.
Thence, a hidden agenda from post 2015 reviews of sdgs. At 4 UN NY and Geneva events. I saw a clear and present conflict between older and younger halves of the world. Elders have not yet made system transformation changes sustainable millennials would need to money, education, energy to name but three.
Regarding process design, a short summary of real intelligence problems could be input by Guterres top team to the start of project.
Research for Von Neumann’s biography shows: peak mathematical genius is rare. Fei-Fei Li could be asked with Hassabis to form a maths subgroup to review deep human intelligence challenges that UN’s unique purpose in the world serves in this system transformation decade
NB a striking coincidence of Neumann’s life and times. His 12 years of life after World War 2 involved a struggle of everything his triad (Einstein Turing) wanted to open source to the world that the top secret era had previously constrained. From chaos maths, he saw that computing as engineers’ sixth compound multiplier would spin a worldwide singularity crisis requiring extraordinary finesse and trustworthy behaviors at critical times these 2020s are evidencing.