MLE | Data Science | TrueML | Ex-Amazon | Built models empowering 1% of video ads in the U.S.
New York City
Joined July 2014
- Tweets 1,124
- Following 174
- Followers 245
- Likes 6,007
Pinned Tweet
Ever wonder how to make GenAI smarter?
The short answer: make it larger.
But how, and why does that translate into billions of dollars spent on AI infrastructure? I break it down in my new article.
If you’re interested, write “article” in the comments and I’ll DM it to you.
Plot twist: @ylecun was a member of a secret society of scientist trying to halt progress towards AGI. They have members in all major companies and research labs, but he was the most successful one.
I'm curious to know which points you agree or disagree with. If you found this post insightful, please share it so others can see and add their comments as well.🙏
Full post: hdarabi.com/index.php/2024/0…
I'm not an economist, and these are purely my own reflections. No major decisions, whether career-related or otherwise, should be made based solely on this perspective, and I do not assume any responsibility for such decisions.
Lastly, we're likely to witness a mindless rush towards AI by many companies. There will be a significant demand for individuals with LLM-related projects or publications listed on their resumes over the next five years.
The demand for senior professionals, who can complement AI capabilities, will increase, and so will the standards for reaching such levels. Traditional skills like leet coding might become less relevant, whereas expertise in architecture, statistics, and modeling will gain value.
* In a competitive market, most advantages will ultimately benefit the end-user through improved features that are more accessible.
Another notable trend is the widening gap between senior and juniors in data science and software engineering.
* A small part of the economic benefit goes to data scientists with their enhanced capacity to achieve more.
* A larger share will benefit company owners to utilize the same number of data scientists to accomplish more, but
If we assume these tasks currently account for about half of our workload, and we can now perform them twice as fast, it's straightforward to project a productivity gain of 33%. How will this benefit be distributed?
Therefore, I believe the primary beneficiaries of LLM advancements will be professional data scientists.
* Train models and perform inference.
On the flip side, the need to:
* Write repetitive code, especially for data wrangling, and
* Produce extensive documentation, reports, and visualizations,
will significantly decrease.
My first point is that LLMs will indeed simplify coding and modeling tasks, but primarily for those who are already professionals in the field. You still need to possess the skills to:
* Translate a business problem into a mathematical model,
* Identify relevant data, and
Now that we've moved beyond the initial ChatGPT frenzy, with claims that all tech jobs are at risk due to LLMs automating code and model generation, I'd like to share my thoughts on how our roles as data scientists and software engineers might evolve in the near future.
If you're curious about a straightforward example, feel free to reach out via direct message. Every other day, I share insights like this. Follow me for more content. Thank you!🙏