“Keeping Humans in the Loop” is about non-hype AI/ML and tech news that I deem worthy of sharing.
You can find out more about me on www.taranrai.me
You may also find the Q and A below helpful:
Who is your target audience? Anyone who wants to read my posts. I don’t discriminate.
What kind of topics will you cover? Mainly AI/ML and tech topics. There will be some posts on AI in veterinary and human healthcare. But, there will also be AI shitposts. I am not too serious about this.
Will you cover topics other than AI/ML? Yep. Some of the posts will be on general tech or even politics. Basically, anything I find interesting to share.
What if I don’t understand your posts? Google, ChatGPT and other chat bot Large Language Models (LLMs) are your friends. Although, do let me know what bits are confusing so I can work on this too.
How often will you post? It depends on whether I have anything valuable to say. I don’t plan on bombarding your inbox with annoying trivial updates.
Why use the term AI/ML? For several years I worked on what was considered Machine Learning (ML) applied mainly to the digital pathology space and some business problems. Then early 2023 I went on a month-long holiday and came back to people labelling everything I did as “AI”. I'm not fully comfortable with this shift of terminology but I’ve accepted it for now. ML is focused on specific data-driven methods, which can be considered a subset of AI, which in turn broadly simulates human intelligence, including reasoning. Recently, Large Language Models (LLMs) have made significant advancements. However, in my opinion, we're still not at the point of fully solving Artificial General Intelligence (AGI) (whatever that means), and it might require a new AI paradigm to achieve that impressive milestone.
What is your educational background? I have a PhD applying AI/ML to digital pathology, an MSc in data science and a BA (Dual Hons) in economics and finance.
Does your experience in AI/ML give you more expertise in these domains compared to those without similar academic qualifications? Absolutely not. There will be some people in the field for only a year or two, who have no formal education but may have made great strides in understanding and developing engineering practices to create “AI” software and products. Tbh everyone can form an opinion or contribute to this field. I’m not a fan of gatekeeping and think that those who enforce these practices can fuck off. This Threads post pretty much captures this sentiment.
Why did you apply your original AI/ML research to veterinary rather than human digital pathology? Because of the concept of One Health, One Medicine: where the interconnectedness of human health and animal health is viewed as a unified entity. AI/ML breakthroughs in veterinary medicine can lead to advancements that benefit both humans and animals. I also like dogs.
Do you think of AI as an existential threat? No, but it doesn’t mean I don’t have any concerns about the current state of AI/ML. We are dealing with models that, if not developed
or regulated1responsibly, could lead to unintended (more immediate) consequences. Also related, I have many questions when it comes to “AI alignment”, for example, alignment with who?Should I take your views seriously? Nah, I’m just as lost as everyone else.
Will you make mistakes? Yes, I’m only human.
Are you planning to make lots of money from these blog posts? No. Not yet anyways.
So why bother with writing posts? In all honesty, it’s more for me than for you. I'm not capable of absorbing and retaining every bit of info I come across. I sometimes use writing as a method to process and organise information, basically creating a personal archive I can refer to in the future. If you also benefit from these posts then great.
Can any of the answers to the above questions change? Yes, of course.
Can your views on AI/ML and any of your posts change? Yes, absolutely. When presented with new information, it’s not uncommon to change views or pivot to other ideas. It’s unscientific (and stupid) to fervently defend something that has proven to be wrong.
Not all of your posts come through my email inbox, why? Correct, there will be times where I’ll post articles that are a bit too niche where they’re not relevant for everyone. The emailed newsletter-blog type posts are for a more general audience who have an interest in AI and tech.
If you liked this and want to read my other posts then please subscribe.
For additional info on me, check out my personal website: www.taranrai.me
My belief since June 2024 is that we’re overregulated in some domains and believe this to be especially true in the UK and EU. I’m not against regulation so let me make that clear. We need it and if we don’t have any, we would be seriously harmed ranging from economic instability to public health crises. Nevertheless, when a technology like AI/ML emerges, finding the right regulatory balance is really difficult and most ideas don’t seem to be sensible or focus on the issues. For AI/ML we need to be careful not to stifle innovation and defo not create further barrier to entries for smaller players to get involved.
