AI in Africa fueled by what would have been tax dollars (Bill Gates as Upgraded-with-LLM White Savior)

Let’s check in on what would have been our tax dollars (capital gains tax on appreciated Microsoft stock) if not for our wise policy of allowing rich douches to deduct unlimited amounts for “charity” (not to be confused with “vanity”!)… “Gates Foundation Pledges $1 Billion to Combat A.I. Inequality” (New York Times, today):

The foundation’s annual Goalkeepers report, released on Monday, advocated an urgent effort to ensure that A.I. “helps narrow gaps between the richest and poorest rather than widening them.”

The foundation said that it would divide the new funding among external groups working on its priorities, spanning work like encouraging A.I. use among doctors, farmers and teachers, or developing data sets in more languages.

Last month, Mr. Gates warned that A.I. would spread across the economy, possibly causing job losses across the economy and leaving little room for one industry to absorb the refugees from another. The biggest tech companies in the world, including Microsoft, are racing to help corporate customers adopt A.I., and have said it could replace many workers.

The visionary was telling Americans to bulk up on low-skill immigrants just a year before Google kicked off the LLM revolution by publishing its transformer architecture (BBC 2016):

The billionaire philanthropist Bill Gates has told the BBC that the United States “should set a better example” by taking in more refugees.

Mr Gates said his home country “had the capacity” to follow the examples of Germany and Sweden, who were “to be congratulated” for welcoming migrants.

Speaking at the World Economic Forum in Davos, Mr Gates added that governments were dealing with “tight budgets”. However, he emphasised that the issue was largely a political one, and that “the total number of refugees is not a world record”.

The statement about “tight budgets” is confusing. We’re informed that low-skill migrants make a society richer. If true, why is a government budget a limiting factor? Separately, how has Germany done economically and culturally since 2016 thanks to its virtuous refugee policy? The refugees of 2016 are now fully established in the German labor market. With unprecedented levels of low-skill immigrant residents, Germany should thus be experiencing an unprecedented economic boom. The Wikipedia page titled “German economic crisis (2022–present)” says “[Germany] became the worst-performing major economy globally in 2023” (attributed to “slowing immigration to Germany” because, of course, the fix when open borders doesn’t work is more-open borders).

How about Sweden? With government at 50% of GDP, the country was roughly evenly balanced between those who benefit from a larger government and those who are harmed by an expansion of government, welfare, etc. (The U.S. percentage of GDP devoted to government is similar if we include nominally “private” (but heavily regulated) health care.) In a recent narrowly-decided election, the balance was tipped to the “expand the government and keep the borders open” party by immigrant voters (about 13% of those eligible; source; see also this X post).

Was Gates’s 2016 statement due to the crisp mountain air at Davos? No.

the U.S. immigration laws are bad – really, really bad. I’d say treatment of immigrants is one of the greatest injustices done in our government’s name. (Bill Gates, quoted in Rolling Stone 2014)

Why do Africans need a snow white savior and 7 dwarf NGOs to tell them how to use ChatGPT? Is this new Gates Foundation program an example of epistemic injustice? (Wikipedia: “Epistemic injustice occurs when an individual or group is wronged in their “capacity as a knower”, meaning that their ability to produce knowledge is called into question.”) Here’s a still frame from Bill Gates’s video:

ChatGPT says that it is in the Wolof language, spoken in “Senegal and also in The Gambia and Mauritania”. ChatGPT adds “I can usually read, translate, summarize, and explain Wolof documents, especially standard written Wolof.” So the video example of how AI doesn’t work for African languages is provided in a language that the most familiar AI understands well enough to translate.

(ChatGPT identifies the photo as likely from Saloum, Senegal and can give some history: “religiously motivated Muslim violence and coercive Islamization occurred in Saloum, especially during Maba Diakhou Bâ’s jihad of the 1860s”; it was a peaceful jihad according to George W. Bush that resulted in Muslims becoming 97 percent of the population in Senegal (state.gov))

Readers: since you’ll have to pick up the slack and pay the taxes that Bill Gates skipped on via his foundation, what do you think of this program? If Gates Foundation employees aren’t going to work at a company training and/or running an LLM how are they going to have a significant effect on what AI can do for the Africans that the Gates Foundation purports to understand? Unless they think that Africans are stupid, wouldn’t it make more sense just to give $1 billion to Africans and let them spend it on whatever they believe that they need? Even if having trained do-gooders spend money rather than letting the purported beneficiaries choose, is the local language idea a good one? A lot of people get motivated to learn widely-spoken languages because proficiency will enable them to play a desired videogame or watch and understand a movie (estimate of at least 200 million English learners for video gaming in 2020). Imagine the motivation for someone in Senegal to learn French or English if that were the only way to take advantage of AI (which, of course, it currently isn’t since ChatGPT actually does understand Wolof and 80 percent of Senegal’s adults already have smartphones).

Finally, if AI is unsafe, which Bill Gates says it is, why would working to make an unsafe technology more widespread in Africa, or anywhere else, be a noble endeavor?

A related classic from the pre-AI era… “For God’s Sake, Please Stop the Aid!” (Der Spiegel):

The countries that have collected the most development aid are also the ones that are in the worst shape. Despite the billions that have poured in to Africa, the continent remains poor. Huge bureaucracies are financed (with the aid money), corruption and complacency are promoted, Africans are taught to be beggars and not to be independent. In addition, development aid weakens the local markets everywhere and dampens the spirit of entrepreneurship that we so desperately need. As absurd as it may sound: Development aid is one of the reasons for Africa’s problems. If the West were to cancel these payments, normal Africans wouldn’t even notice. Only the functionaries would be hard hit. Which is why they maintain that the world would stop turning without this development aid.

A portion of the corn often goes directly into the hands of unsrupulous politicians who then pass it on to their own tribe to boost their next election campaign. Another portion of the shipment ends up on the black market where the corn is dumped at extremely low prices. Local farmers may as well put down their hoes right away; no one can compete with the UN’s World Food Program.

… believe me, Africa existed before you Europeans came along. And we didn’t do all that poorly either.

Loosely related, my latest way of responding to anyone in academia who expresses an interest in AI safety, AI regulation (now that Trump is anti-regulation, all of my academic friends are very keen on government-imposed restrictions), etc.:

I do feel unsafe. Current models are close to PhD-level intelligence. At some point in 2027, they’re expected to achieve the intelligence of a human who decided not to enter a PhD program.

Not very related, my X response to a New York Times article about the horror of Sydney Sweeney being shown to the general public:

7 thoughts on “AI in Africa fueled by what would have been tax dollars (Bill Gates as Upgraded-with-LLM White Savior)”

  1. Did the Gates Foundation consult Dr. Prof. Uju Anya, the prominent Nigerian linguist? She offered highly specific, actionable advice to white men like Mr. Gates.

    • Anon: I’m not sure that I understand the craze for Sydney Sweeney. Maybe it is my advanced age, but I think Audrey Hepburn, Grace Kelly, and various 1970s and 1980s supermodels were more attractive in their respective days. On the other hand, by devoting (virtual) ink to this issue, the New York Times seems to be demonstrating that they don’t believe their own rhetoric. Earth is melting due to Donald Trump-caused climate change. U.S. democracy is ending because of Donald Trump and various other Nazis, including Elon Musk. Also, people in this doomed country on this doomed planet should pay attention to which actress or model appears in an advertisement.

  2. Many of your readers (myself included) are not technology experts like you, so the insights you and your network provide are quite interesting. With that in mind, here is an idea for a post on AI involving current events, and hopefully you could leverage your network of tech-minded friends to chime in too (is there unanimity among them? What are the major debates etc.). Broadly, if you could address:

    1) Is AI on the cusp of becoming a danger to humanity?
    2) Have the recent warnings by Anthropic been cooked up/amplified to achieve regulatory capture and/or maintain what appears to be a near duopoly with OpenAI?
    3) Is Jacob Coxon a “plant” by Anthropic? He appears to be a rather junior employeed that not many had heard of until now?
    4) Is it a coincidence that many others are instantly piling on to Coxon’s “warning” (e.g. Barack Hussein Obama, who of course is an accomplished expert in technology)?
    5) Any other observations that would be useful to think about for those of us who are not steeped in technology?

    • Thanks for the vote of confidence. I was just out in Silicon Valley for a friend’s wedding. A couple of the other people there were top managers at enormous enterprises that feed the AI engines we love so much (they didn’t work directly on an LLM, but rather produced and/or managed the infrastructure that is necessary for LLMs). When these guys refer to the CEO of Nvidia as “Jensen” it is because they actually interact with him on a routine basis. I figured that they’d be much better-equipped to predict where we’d be in a year or two than I was. The surprise was that they admitted to having no clue as to what was going to happen, good or bad, with AI! So if the folks who really have their “ears to the rack” (instead of “ears to the ground”?) have no clue, what hope is there for the rest of us.

      That said, your idea for a post makes sense. Anthropic is especially noxious due to its association with Effective Altruism, Sam Bankman-Fried, Caroline Ellison, et al. “I would rather be governed by the first 2,000 people in the Boston telephone directory than by the 2,000 people on the faculty of Harvard University,” was William F. Buckley, Jr.’s famous observation. Maybe we could update this for our time: “I would rather be governed by 2,000 randomly selected colleagues of Jason Arday and Claudine Gay than by the 2,000 leading Effective Altruists.”

    • Okay, so your quick response, particularly relative to discussions with fellow tech experts in CA, here was actually incredibly interesting from my uninformed perspective!

      Please do follow up with more in a full post (ideally, multiple posts as things evolve…appears to be very fast-moving). FYI, you may believe that your understanding of what is going on (like your interactions with folks in CA) is well understood by most people. I can assure it is not! A lot of us are dying (I know, bad word) to understand what is really going on, so please don’t refrain from describing details that you may consider banal. This story is obviously very interesting and most of the coverage I see in the major “newspapers” is uniformed.

  3. “80 percent of Senegal’s adults already have smartphones (actually, obamaphones funded by US taxpayers)”

Leave a Reply

Your email address will not be published. Required fields are marked *