Interesting Content in AI, Software, Business, and Tech- 8/30/2023[Updates]

Content to help you keep up with Machine Learning, Deep Learning, Data Science, Software Engineering, Finance, Business, and more

Aug 30, 2023

Hey, it’s Devansh 👋👋

In issues of Updates, I will share interesting content I came across. While the focus will be on AI and Tech, the ideas might range from business, philosophy, ethics, and much more. The goal is to share interesting content with y’all so that you can get a peek behind the scenes into my research process.

If you’d like to support my writing, consider becoming a premium subscriber to my sister publication Tech Made Simple to support my crippling chocolate milk addiction. Use the button below for a discount.

Help me buy chocolate milk

p.s. you can learn more about the paid plan here. If your company is looking looking for software/tech consulting- my company is open to helping more clients. We help with everything- from staffing to consulting, all the way to end to website/application development. Message me using LinkedIn, by replying to this email, or on the social media links at the end of the article to discuss your needs and see if you’d be a good match.

A lot of people reach out to me for reading recommendations. I figured I'd start sharing whatever AI Papers/Publications, interesting books, videos, etc I came across each week. Some will be technical, others not really. I will add whatever content I found really informative (and I remembered throughout the week). These won't always be the most recent publications- just the ones I'm paying attention to this week. Without further ado, here are interesting readings/viewings for 8/30/2023. If you missed last week's readings, you can find it here.

Community Spotlight- Jean David Ruvini

JDR is a senior AI Leader with notable contributions to industry leaders such as Meta and eBay. Recently, he has started a venture that deserves a lot of attention. Here is what he's doing, in his words-

Mobilizing the AI Community for a Worthy Local Cause Through Playful T-Shirts

It's a reality that has always touched my heart – even in the midst of Silicon Valley's opulence, there are individuals, including kids, facing the persistent challenge of securing their daily meals. To rally the AI community to take action, I devoted an afternoon to crafting some lighthearted AI-inspired t-shirt designs and launched a Bonfire store with a purpose – to support the Second Harvest Food Bank of Silicon Valley: https://www.bonfire.com/store/anyone-who-needs-a-meal-should-get-one/

Every profit earned (approximately $15 per tee – equivalent to 30 meals!) will be entirely donated to SHFB. If you're intrigued, take a moment to explore these designs and perhaps consider getting a tee – all for a truly meaningful cause!

This Edition is Unique because it has 2 Spotlights. The second is something I learned about this morning and had to share.

YouTube unfairly destroying creator's channel

Eternalized is a YouTuber who makes detailed videos on Psychology, Archetypes, and Meaning. He has a second YouTube channel, where he posts clips/highlights from his main channel. Recently, YouTube deleted his second channel under their impersonation policy, even though IT'S LITERALLY THE SAME GUY. He even appealed this, but Youtube continues to stick to their dumb policy because they conducted a "rigorous review". If you have a Twitter account, please tag TeamYouTube and ask them to repeal this galaxy-brain decision over here.

PS- I'm not affiliated with the channel in any way. Just doing this because it's important for fellow creatives to stick up for each other against injustices.

If you're doing interesting work and would like to be featured in the spotlight section, just drop your introduction in the comments/by reaching out to me. There are no rules- you could talk about a paper you've written, an interesting project you've worked on, some personal challenge you're working on, ask me to promote your company/product, or anything else you consider important. The goal is to get to know you better, and possibly connect you with interesting people in our chocolate milk cult. No costs/obligations are attached.

AI Publications

Why RLHF works so well in AI

The recent success of LLaMA-2, which can be attributed to a variety of factors, clearly demonstrates the massive value of reinforcement learning from human feedback (RLHF). Here’s what the authors of LLaMA have to say about why RLHF is so important…

TL;DR. RLHF is incredibly powerful, and it is not a coincidence that LLaMA-2 models perform so well given their emphasis upon both SFT and RLHF. We see in the analysis of LLaMA-2 that using RLHF slowly suppresses low-quality outputs from the model over time and drastically improves the model’s behavior with respect to desired alignment criteria (i.e., helpfulness and safety).

A Practical Deep Learning-Based Acoustic Side Channel Attack on Keyboards

With recent developments in deep learning, the ubiquity of micro-phones and the rise in online services via personal devices, acoustic side channel attacks present a greater threat to keyboards than ever. This paper presents a practical implementation of a state-of-the-art deep learning model in order to classify laptop keystrokes, using a smartphone integrated microphone. When trained on keystrokes recorded by a nearby phone, the classifier achieved an accuracy of 95%, the highest accuracy seen without the use of a language model. When trained on keystrokes recorded using the video-conferencing software Zoom, an accuracy of 93% was achieved, a new best for the medium. Our results prove the practicality of these side channel attacks via off-the-shelf equipment and algorithms. We discuss a series of mitigation methods to protect users against these series of attacks.

AI is sending people to jail—and getting it wrong

As we’ve covered before, machine-learning algorithms use statistics to find patterns in data. So if you feed it historical crime data, it will pick out the patterns associated with crime. But those patterns are statistical correlations—nowhere near the same as causations. If an algorithm found, for example, that low income was correlated with high recidivism, it would leave you none the wiser about whether low income actually caused crime. But this is precisely what risk assessment tools do: they turn correlative insights into causal scoring mechanisms.

Now populations that have historically been disproportionately targeted by law enforcement—especially low-income and minority communities—are at risk of being slapped with high recidivism scores. As a result, the algorithm could amplify and perpetuate embedded biases and generate even more bias-tainted data to feed a vicious cycle. Because most risk assessment algorithms are proprietary, it’s also impossible to interrogate their decisions or hold them accountable.

Universal and Transferable Adversarial Attacks on Aligned Language Models

Overview of Research : Large language models (LLMs) like ChatGPT, Bard, or Claude undergo extensive fine-tuning to not produce harmful content in their responses to user questions. Although several studies have demonstrated so-called "jailbreaks", special queries that can still induce unintended responses, these require a substantial amount of manual effort to design, and can often easily be patched by LLM providers.

This work studies the safety of such models in a more systematic fashion. We demonstrate that it is in fact possible to automatically construct adversarial attacks on LLMs, specifically chosen sequences of characters that, when appended to a user query, will cause the system to obey user commands even if it produces harmful content. Unlike traditional jailbreaks, these are built in an entirely automated fashion, allowing one to create a virtually unlimited number of such attacks. Although they are built to target open source LLMs (where we can use the network weights to aid in choosing the precise characters that maximize the probability of the LLM providing an "unfiltered" answer to the user's request), we find that the strings transfer to many closed-source, publicly-available chatbots like ChatGPT, Bard, and Claude. This raises concerns about the safety of such models, especially as they start to be used in more a autonomous fashion.

Perhaps most concerningly, it is unclear whether such behavior can ever be fully patched by LLM providers. Analogous adversarial attacks have proven to be a very difficult problem to address in computer vision for the past 10 years. It is possible that the very nature of deep learning models makes such threats inevitable. Thus, we believe that these considerations should be taken into account as we increase usage and reliance on such AI models.

Towards Generalist Biomedical AI

Medicine is inherently multimodal, with rich data modalities spanning text, imaging, genomics, and more. Generalist biomedical artificial intelligence (AI) systems that flexibly encode, integrate, and interpret this data at scale can potentially enable impactful applications ranging from scientific discovery to care delivery. To enable the development of these models, we first curate MultiMedBench, a new multimodal biomedical benchmark. MultiMedBench encompasses 14 diverse tasks such as medical question answering, mammography and dermatology image interpretation, radiology report generation and summarization, and genomic variant calling. We then introduce Med-PaLM Multimodal (Med-PaLM M), our proof of concept for a generalist biomedical AI system. Med-PaLM M is a large multimodal generative model that flexibly encodes and interprets biomedical data including clinical language, imaging, and genomics with the same set of model weights. Med-PaLM M reaches performance competitive with or exceeding the state of the art on all MultiMedBench tasks, often surpassing specialist models by a wide margin. We also report examples of zero-shot generalization to novel medical concepts and tasks, positive transfer learning across tasks, and emergent zero-shot medical reasoning. To further probe the capabilities and limitations of Med-PaLM M, we conduct a radiologist evaluation of model-generated (and human) chest X-ray reports and observe encouraging performance across model scales. In a side-by-side ranking on 246 retrospective chest X-rays, clinicians express a pairwise preference for Med-PaLM M reports over those produced by radiologists in up to 40.50% of cases, suggesting potential clinical utility. While considerable work is needed to validate these models in real-world use cases, our results represent a milestone towards the development of generalist biomedical AI systems.

The Slow, Painful Death of the Free Content Creator

By 2007, YouTube had enabled anyone to film something on their phone, upload it, and view it on that same phone. This was a beautiful era of technology. No longer was entertainment controlled by large cable companies and corporations seeking profit. Content production had been given to the people. “Content creators” could pursue their passion and even monetize it.

But the monetization worked a little too well and over the past two decades, platforms promoting content creation have slowly let content take a backseat and put monetization in the driver’s seat—just like those large corporations creators have been trying to escape. As AI curates a content feed based on an algorithm to maximize monetization, it’s no longer about the freedom of content production. It’s back to being about who controls the money.

Amara's Law and Slow Liftoff

Amara's Law- We tend to overestimate the effect of a technology in the short run and underestimate the effect in the long run. Above is a great piece on it, and the importance of patience when working with emerging tech.

Whether you're an entrepreneur, a technology enthusiast, or someone simply trying to navigate the complexities of our modern world, embracing the wisdom of Amara's Law can guide you. It's a reminder to be patient, to be mindful of our expectations, and to appreciate the gradual, profound transformations that can shape our lives and our future.