MyView: We are clearly far from #ChatGPT understanding causal relationship- especially more complex relationships.
However, you may be able to work with ChatGPT to think through parts of #Causal model / DAG.
#Causality#CausalAI @CausalInferBot #CausalTwitter
@PHuenermund @yudapearl@jhurwitz@GaryMarcus#ChatGPT understands data as well as a young child understands the theory of special relativity. A child might say "E=mc²" but they don't "understand" the meaning.
Chat GPT certainly read @yudapearl's Book of Why and can do an express job of regurgitating info!
@CausalInferBot
@CraigMilroy I was recently talking to an engineer at a startup #Biotech company whose company had a "lunch & learn" session on #GPT4 uses.
A colleague had never used it. After the lunch she used ChatGPT to help kickstart the design of a study that she had been struggling with for weeks.
MyView: #Microsoft is doing an impressive job of quickly infusing #GPT4 into its existing offerings. It sometimes takes teams multiple quarters or years to integrate emerging tech into existing offerings.
Impressive #DevOps practices behind the scenes.
#AI#MachineLearning
Major new #AI announcement today:
@GitHub#Copilot X, which will bring the impressive power of @OpenAI's new #GPT4 to coders.
#CEO@ashtom personally walks us through the big news, which is extensive + will bring *major* boosts in developer productivity.
The details:
1/x 🧵
@CraigMilroy If I can plug a book I co-authored...
#AugmentedIntelligence: The power of Human Machine Collaboration
The premise: Human+Machine is more powerful than either of them on their own.
https://t.co/gTm10gNkdW
@CraigMilroy Agree with you @CraigMilroy.
Although most of the press about #GPT and #GenerativeAI has been focused on #AI destroying jobs, businesses need to focused on #AugmentedIntelligence.
Meaning, how can employees exploit the power of #GPT4 in combination with human intelligence.
The future of #AI is #Causality.
🧐 3 things you need to know now!
> Collaborating through visual models
> Abstracting the complexities of #OpenSource#Causal libraries
> How #CausalAI is delivering value today
@CausalInferBot #CausalTwitter#GartnerDA
https://t.co/34JHsFDVBm
Fundamental problems with #Data and #AI approaches that require you to rethink your data stack:
-Competitors moving quickly
-Lack of internal skill
-Budget constraints
#CausalAI allows you to focus on specific challenges and gathering relevant data. #Gartnerda @CausalInferBot
We live in a world where data products don't just exist independently. Maybe an unpopular opinion, I will say it. Data Mesh is interesting, but I'm not convinced most organizations have achieved the maturity to enable it. #GartnerDA
@CraigMilroy Start by identifying your business challenge and desired outcomes.
Collecting all available data leads to "noise" and poor results.
Instead, strategically collect the right data to understand causal relationships for better #AI. That's where #CausalAI comes in. #GartnerDA
👉More data≠Better #AI 👈
MyView: There is a fallacy that if you can just get enough data your #ML models will be smarter. Instead you need to think about the relationships between variables and outcomes.
Hence, #CausalAI is the next big thing in AI
#gartnerda#causaltwitter
MyView: There needs to be a business case for the #cloud. If your goal is agility, #CloudComputing is a great platform for spinning up new ideas and testing POCs.
Once you understand compute, storage and networking needs, it may make sense to bring back on premises. #HybridCloud
MyView: Fact
Traditional #AI is based on correlations while #CausalAI helps to understands the causal relationships between variables and outcomes.
#Gartnerda#CausalInference#DataScience
Upcoming Webinar:
https://t.co/Y2MBKmggDT
To deliver meaningful output from #data you need to understand the context of data, the business problem your trying to solve and how the business process works. #CausalAI#Causality
Quant methods hot take: If you lack the domain knowledge required to draw the relevant DAG or set meaningful priors, you should neither fit nor attempt to interpret the model. Back away slowly from your stats software and do a comprehensive lit review.
@tcrawford@united@Zoom These types of partnerships would have been great at #zoomtopia! If #zoom was focused on the enterprise they would build a platform that wasn't blocked by airlines, hotels, etc.
How many articles start with predicting the end of the #mainframe?
The #cloud is the de facto standard for many startups, but on-premises will persist forever. When larger organizations examine #cloudcosts and modernization, status quo makes sense for existing apps.
August 2019: Gartner predicts that by 2025, 80% of enterprises will shut down their traditional data centers. In fact, 10% of them already have.
Jan 28, 2022: More than half (51%) said they planned to close all their traditional data centers in the next 24 months.
1/2