Tech majority disagrees with AI warnings from Hawkings, Musk and Gates

Arizona State researchers used machine learning to analyze tweets about artificial intelligence and found most people in the tech community are optimistic about AI compared with AI experts

Tech majority disagrees with AI warnings from Hawkings, Musk and Gates

Tech star personalities Stephen Hawkings, Elon Musk and Bill Gates warned the public about artificial intelligence (AI). The tech-oriented public and AI experts disagree, though, according to a recent research paper, “Tweeting AI: Perceptions of AI-Tweeters (AIT) vs Expert AI-Tweeters (EAIT),” (pdf) published by researchers at the School of Computing, Informatics and Decision Systems Engineering at Arizona State University.

 One of the insights from this work, extracted from the tweets analyzed:

“Co-occurring patterns tell us that AIT are in general fantasizing about the future whereas EAIT are grounded and realistic.”

Study authors used statistical analysis, sentiment analysis and machine learning to learn this insight and summarize the study with the conclusions below.

1. Tweets about AI are overall more positive compared to the general tweets. Despite the overall negative sentiment of Twitter, overall the 2.3 million tweets analyzed about AI are positive by a large margin. The researchers found 65 percent positive sentiment of this sample from the Twitter firehose compared to the negative overall negative Twitter sentiment reported by Microsoft and Cornell researchers in a 2011 Association for Computing Machinery research paper.

2. Tweets posted by AI experts are more negative than the general AI tweeters, which without the statistics support the common folklore that tech fan opinions as unwarrantedly optimistic compared to evidence-based opinions. The distribution of expert tweets was more evidence-based, with more than 50 percent about the impacts of AI and research directions in the field.

3. Tweets posted by experts have lower diffusion than the tweets posted by general AI tweeters. The general AI tweeters averaged 2 retweets and the experts near 0.

4. General AI tweeters are geographically distributed with London (6.3 percent), New York City (4 percent), Paris (2.9 percent) and San Francisco (2.5 percent). Experts, to no one’s surprise, follow the money and are concentrated in San Francisco (10 percent) and Seattle (6.6 percdent), which reflects the long-term investment in AI by Google and Facebook in the San Francisco area and Microsoft in the Seattle area. A chart of the locations of both groups is included below.

ai tweets user locations Arizona State University. Manikonda, L.; Dudley, C.; and Kambhampati. 2017.

This study corresponds to the opinion of a Harvard Business Review study based on a survey of individuals and a Stanford and Microsoft longitudinal study of AI-related stories in The New York Times between 1986 and 2016 (pdf) that revealed the discussion has sharply increased in optimism beginning in 2009.

How the tweeting AI study was performed

The researchers started by crawling Twitter and extracted 2 million unique tweets with the hashtags #ai and #artificialintelligence to build a larger list of co-occurring hashtags. The top four hashtags #ai, #artificialintelligence, #machinelearning and #bigdata and the list of most frequent tweeters were to extract 2.3 million unique tweets as the study data set.

A list of the most influential AI experts was compiled by IBM based on thought leaders who are active on Twitter, speak regularly at conferences and produce AI research and entrepreneurs working on interesting related products. IBM also interviewed experts to complete the list of the most influential AI experts on Twitter below.

  • Nathan Benaich
  • Joanna Bryson
  • Alex Champandard
  • Soumith Chintala
  • Adam Coates
  • The CyberCode Twins
  • Jana Eggers
  • Oren Etzioni
  • Martin Ford
  • Adam Geitgey
  • John C. Havens
  • J.J. Kardwell
  • David Kenny
  • Tessa Lau
  • Dr. Angelica Lim
  • Staya Mallick
  • Gary Marcus
  • Chris Messina
  • Elon Musk
  • Dr. Andy Pardoe
  • Alec Radford
  • Delip Rao
  • Dr. Roman Yampolskiy
  • Gideon Rosenblatt
  • Adam Rutherford
  • Matt Schlicht
  • Murray Shanahan
  • Amir Shevat
  • Rest Sidhu
  • Adelyn Zhou

Using a machine learning-based unigram analysis of the profiles of the user profiles the occupation of the artificial intelligence tweeters (AIT) and the expert artificial intelligence tweeters (EAIT) were categorized.

ai tweeters occupations Arizona State University. Manikonda, L.; Dudley, C.; and Kambhampati. 2017.

The top 100 tweets of each user were categorized using a machine learning statistical model to find similarity of interests within each group. AIT users interests were categorizes as

AIT Users’ interests

Science and technology

Personal status and opinions

General news

Non-English tweets

Daily updates

EAIT users’ interests were categorized as:

AIT Users’ interests

Rise of AI

AI subscriptions and research

AI news from industry

Marketing & AI

Opinions about AI

Optimism was defined as being hopeful and pessimism defined as tending to see or believe the worst will happen. The sentiment of the words in the 2.3 million tweets was determined using a popular psycho-linguistic tool proven to identify the emotions in the usage of language.

A machine learning model called Word2Vec2 was trained to understand the linguistic contexts of co-occurring concepts represented by the words in the tweets to categorize the positive and negative sentiment in the corpus of tweets.

This story makes two broad points, one about machine learning and the other about Twitter. Machine learning was shown to do the work that people and less-detailed analytic scoring methods once did in understanding sentiment from a large corpus of people’s communications. Tweets need to be interpreted with the understanding that in general, the sentiment is negative, but groups such as the AIT that fantasize about the future can skew opinion and discretion should be applied to tweets before they are believed.

Join the Network World communities on Facebook and LinkedIn to comment on topics that are top of mind.

Copyright © 2017 IDG Communications, Inc.