
UT Austin McCombs Information Management Program | Business Technology Research and Learning | MIS BBA | Follow us for #Tech Updates & Insights | @UTexasMcCombs
Austin, TX
Joined December 2011
- Tweets702
- Following609
- Followers401
- Likes1.1K
The UT Information Management (IM) Program is a part of The University Texas McCombs School of Business
xitter.cf/i/status/1193614258153…
A new wave of Texas McCombs leaders are shaping the future of business. Our dynamic programs are featured nationally! Check your local listings for #SuccessFiles.
1
8
UT IM Program retweeted
After delivering 72 spacecraft to orbit, Falcon 9 returns to Earth and completes SpaceX’s 200th landing of an orbital class rocket
UT IM Program retweeted
Great experience speaking about "The Future of Generative AI in Education" with Scott Klososky, Robyn Hrivnatz, Kathy Wetzel, and Ashton Lux in San Antonio and Fort Worth at the TASB Summer Leadership events
UT IM Program retweeted
UT Austin's John Goodenough, who changed the world with batteries, dies at 100 kut.org/energy-environment/2…
UT IM Program retweeted
I was honored to celebrate Eve Stephens as @UTAustin's new @UTAustinPolice Chief of Police at today's Swearing-In Ceremony. I'm grateful for the courage and empathy that she will bring to this extraordinary role. Congrats and Hook 'em, Eve! 🤘 Read more:
news.utexas.edu/2023/07/05/e…
UT IM Program retweeted
The Big Four accounting firms are thinning their consultant ranks after aggressive hiring in the wake of the pandemic. wsj.com/articles/big-four-ac… via @WSJCFO @markgmaurer
UT IM Program retweeted
Uh-oh
GPT-4 is getting worse over time, not better.
Many people have reported noticing a significant degradation in the quality of the model responses, but so far, it was all anecdotal.
But now we know.
At least one study shows how the June version of GPT-4 is objectively worse than the version released in March on a few tasks.
The team evaluated the models using a dataset of 500 problems where the models had to figure out whether a given integer was prime. In March, GPT-4 answered correctly 488 of these questions. In June, it only got 12 correct answers.
From 97.6% success rate down to 2.4%!
But it gets worse!
The team used Chain-of-Thought to help the model reason:
"Is 17077 a prime number? Think step by step."
Chain-of-Thought is a popular technique that significantly improves answers. Unfortunately, the latest version of GPT-4 did not generate intermediate steps and instead answered incorrectly with a simple "No."
Code generation has also gotten worse.
The team built a dataset with 50 easy problems from LeetCode and measured how many GPT-4 answers ran without any changes.
The March version succeeded in 52% of the problems, but this dropped to a pale 10% using the model from June.
Why is this happening?
We assume that OpenAI pushes changes continuously, but we don't know how the process works and how they evaluate whether the models are improving or regressing.
Rumors suggest they are using several smaller and specialized GPT-4 models that act similarly to a large model but are less expensive to run. When a user asks a question, the system decides which model to send the query to.
Cheaper and faster, but could this new approach be the problem behind the degradation in quality?
In my opinion, this is a red flag for anyone building applications that rely on GPT-4. Having the behavior of an LLM change over time is not acceptable.
Have you noticed any issues when using GPT-4 and ChatGPT lately? Do you think these problems are overblown?
Private Cloud a Compelling Option for CIOs: Insights from New Research via @ConstellationR @dhinchcliffe constellationr.com/blog-news…
Microsoft says commercial use of #fusion produced energy will be available within 5 years as part of new deal with Sam Altman-backed @Helion_Energy
msn.com/en-us/money/technolo…
2
99
UT IM Program retweeted
There's a “disconnect between GDP and well-being, especially when we look at digital goods.” How can we account for the value of things like Google Search or WhatsApp in our economic framework? An insightful episode of #ForwardThinkingMGI with @avi_collis: mck.co/AviCollis
UT IM Program retweeted
Eight dimensions of design thinking capture important differences in how managers and designers perform their work and make decisions.
From “Can Design Thinking Succeed in Your Organization?” mitsmr.com/3APnFLa
UT IM Program retweeted
Teachers: "AI is a disaster, how am I going to know who is cheating?!"
Students:
UT IM Program retweeted
Congratulations to Herb Miller on receiving the namesake & first @UTexasMcCombs Marketing Department Herbert A. Miller Jr. Professional Faculty Impact Award! I'm proud of the immense contributions he's made throughout his 39 years of teaching. #whymccombs medium.com/texas-mccombs-new…
UT IM Program retweeted
There are 4 basic team cultures, with different emphases on psychological safety and different emphases on intellectual honesty. Each culture supports or discourages learning and innovation to different degrees. sloanreview.mit.edu/article/…
UT IM Program retweeted
Attention project professionals! Don't miss out on the opportunity to expand your project leader toolkit at our Spring 2022 Professional Development Day. Register now: ow.ly/zUmy50NGzzA or ow.ly/iEXS50NGzzI
UT IM Program retweeted
🤘@UTAustin's 40 Hours for the Forty Acres kicks off today!
Help @UTexasMcCombs change the world by giving back to the people, places, and pursuits that mean the most to you. Learn more:
40for40.utexas.edu/giving-da…
#UT40for40 #WhatStartsHere #WhyMcCombs
UT IM Program retweeted
UT System ranks 2nd in the country (after CUNY) for upward mobility, sending more than 3 times as many students from the bottom fifth of income distribution into the middle class and beyond as all eight Ivy League campuses, plus Duke, M.I.T., Stanford and Chicago, combined!
Company devises ingenious plan to use old #lithium-ion batteries from broken-down #EVs: ‘These batteries work very well’
msn.com/en-us/money/companie…
58
























