DAYMN — 17 Oct 2021
Data Articles You Might Need. Subscribe & follow here!
Here are the top 5 articles from this past week. Please do share your feedback & thoughts on what can be better and if there is anything I can improve!
1. Open Source coming to financial services, finally?
As a big fan of open source and the benefits it brings to developing better common solutions for shared world problems, financial services has been a such strong hold out against making good use of Open Source technologies or contributing to them. Being a recent entrant in this space, I find that a lot of common basic problems are still being solved either with 1970’s tech or with “cutting-edge-fintech-solutions” — with no middle ground for accessible solutions for all. This article from a16z calls this out and the benefits so well!
Call out? Developers will not only be the creators and sellers of these primitives, they will also be the buyers. With a robust ecosystem of open source libraries, product management and development teams will no longer have to seek large budget approval to buy proprietary software from the C-suite. They will be able to experiment with solutions to existing issues, as well as entirely new use cases, for free (or very low cost).
2. Building SEO pages “smartly” using GPT-3
The first time I used GPT-3 was to build a product description page that was BERT friendly and would help us index better on a crawler. GPT-3 has now come such a long way and the descriptions are even more authentic and creative. This wonderful step-by-step article from Mark Grenfell shows one such great example of using GPT-3 to help small and big-time marketers.
https://mackgrenfell.com/blog/building-2-000-unique-seo-pages-with-gpt-3
Call out? Once we had these 2,000 pairs, we built a template page that we could populate with data from each pair. The page’s H1 for example read How to Calculate {metric.name} in {platform.name}. Further down the page we included the description and calculation steps which GPT-3 had written for each of the financial metrics, and also added some information about how Causal integrated with the relevant platform, and how Causal works.
Once all this was done we pushed the changes live and had 2,000 unique SEO pages all ready to go, each targeting a specific use case that customers would have for Causal. You can see an example of the sort of page that was generated here.
3. The essential need to collect both positive and negative signals for your recommendation algos!
Researchers have now proved conclusively that Spotify’s recommendations can get much better, IF THEY HAD A DISLIKE BUTTON. This is such a good vindication of the age-old wisdom, that to build a smart system — we need not only the “I like this data” but also “ I hate this data”. Unless an intelligent system realises both sides of the spectrum, it will keep making inane and innocuous suggestions which are not really useful.
This new article along with the research piece from Cornell is a great read!
https://news.cornell.edu/stories/2021/09/dislike-button-would-improve-spotifys-recommendations
Call out? Specifically, they demonstrated that a listener is roughly 20 percent more likely to “like” a song if the algorithm is trained on 400,000 likes and dislikes, compared to an algorithm trained only on that amount of likes. An algorithm that only has ‘likes’ may help you discover songs that you really enjoy, but it also has a greater chance of recommending songs you don’t like.
4. State of AI Report 2021
Every year for the last 4 years, StateofAI report has been religiously published and ferociously consumed by AI practitioners. This year’s report is another great compilation with the five big themes below.
- Research: Technology breakthroughs and capabilities.
- Talent: Supply, demand and concentration of AI talent.
- Industry: Areas of commercial application for AI and its business impact.
- Politics: Regulation of AI, its economic implications and the emerging geopolitics of AI.
- Predictions: What we believe will happen and a performance review to keep us honest.
Even if you don’t have time to read or skim the 185 page Google-slide, just spend some time on the first 5–7 slides each on Research, Talent and Industry.
Call out? Investors have taken notice, with record funding this year into AI startups, and two first ever IPOs for AI-first drug discovery companies, as well as blockbuster IPOs for data infrastructure and cybersecurity companies that help enterprises retool for the AI-first era.
5. The imminent collapse of Digital Advertising
As someone who has worked a long time in helping marketing teams “spend their dollar better” — I have come to one conclusion, which is never spoken out loudly. A lot of the money pumped into digital ads or “performance marketing” definitely seems to have the great “measurability” component, which makes marketers look good and the powers-that-be satisfied that they have the power to control their marketing messaging, how they want.
However, when COVID hit, I personally witnessed so many companies slash their performance budgets by ~50–60% with absolutely NO IMPACT to their traffic or eventual sales. I’m not claiming that Digital Marketing doesn’t work — it definitely does. What is undeniable though, is the amount of fat and inefficiency in the process that gives the whole approach a bad rap.
This amazing (as usual) article from Scott Galloway shines a light on this!
https://marker.medium.com/the-imminent-collapse-of-digital-advertising-3ab4272be67c
Call out? The basic cheat is the fake view. An ad is reported as being served to humans, when it was actually only “seen” by a bot, or by a person in a “click farm” tapping at dozens of screens, or by nothing at all. Networks of fake websites fool the algorithms into believing they are real publications. Measurements of the impact are all over the map, but we know fraud is pervasive. By one estimate, 88% of digital ad clicks are fake.
Have a wonderful week ahead everyone, hope at least one of these articles is exciting reading material for you, and made you think for a moment
