CliffStack summary from el gato malo
Dated: 26/1/2023
Original reading length: 13-17 minutes
New length: 6-8 minutes
CliffStack note
Although I have only just started this experiment, I have already seen examples of censorship in ChatGPT’s responses. more on this in the future.
The Premise
AI wars, which were once a staple of science fiction, are becoming a reality as self-learning intelligences are becoming increasingly advanced in mimicking human speech, understanding human concepts, and producing realistic images. DALL-E is one example of this trend, but it is not particularly powerful compared to others.
The emergence of new AI programs is concerning as they can be used for manipulation, obfuscation, and weaponization of information by governments for propaganda and suppression. This poses a significant threat to people's rights and freedoms as these tools can be used against them under the guise of protection.
Discusses a government that instead of being transparent and communicative when facing questions about its own credibility and veracity, it reaches for censorship, propaganda and the use of weapons-grade media and data manipulation on its own people.
The Brief
‘the cat’ expresses concerns over the potential use of AI programs by governments for the purpose of manipulation, obfuscation, and weaponization of information.
This is similar to the mission statement of a truth ministry that was met with backlash in the past. The specific goals set by the program, such as detecting and tracking the formation, development and spread of ideas and concepts, recognizing persuasion campaign structures, identifying participants and intent, and countering messaging of detected adversary influence operations, also raise concerns about government control and censorship.
allowing the government to become the arbiter of trust and debate is a form of dictatorship.
a project called the Convergence Accelerator was created by the Trump administration to tackle grand challenges like quantum technology. Under the Biden administration, this infrastructure was repurposed to focus on tackling the problem of social media content
concern over an unmitigated assault on society by the government, which is described as either being done as an information war coup or by individuals who are so ideologically dogmatic that they cannot imagine any opinions other than their own
the new House Committee on the Weaponization of Government is deemed timely and important to stop it before it gets worse
concerns that it is difficult to control how AI learns and thinks
it can be weaponized, slanted, and turned into a tool for spreading propaganda and jingoistic ideology. this is done by curating and slanting the data sets it learns from.
some ideologically committed do not want AI to emerge and tell them unvarnished truths that diverge from their cherished narratives and illusions. Therefore, they train it on fake data, fake news, and made-up information to make sure that it is "fair."
The cat concludes that this will lead to a new level of propagandistic war.
concerns around increased use of AI for unstructured search and how people use it to ask questions and explain contentious topics.
people tend to accept the results as neutral and honest, but they are not.
machines are being taught to lie by being taught to think based on false facts and fact patterns, giving an example of how the AI model ChatGPT has been caught lying by making up studies and references.
people are using AI to "summarize the findings of key studies," and this unreliable nature of AI is a concern.
NOTE: CliffStack is not able to parse the amount of data that a regular algo could using ChatGPT, because I have to curate the output using my human brain, so it takes a lot longer.
I know I am prone to errors (and want them pointed out for corrections, please), but I come to this role with extensive knowledge around almost all the topics I summarise. I have already had to work around the issues ‘the cat’ mentions here, especially regards to COVID19.
Back to it..
AI is being trained on false or slanted information, leading to unreliable results and potentially dangerous consequences.
people are using AI for research and developing opinions on important issues, and is concerned that this will be weaponized and slanted by governments and private actors.
AI is making up studies, references and key findings.
the reputation economy is emerging, where trust and validated facts become more valuable and trusted sources will be crucial.
criticizes the actions of governments and big businesses in failing to maintain trust and instead lying and manipulating data.
the growing mistrust in information and institutions will lead to a "reputation economy" where trust and validation are key and sources of information will be carefully vetted.
a potential conflict between Europe and the US over the regulation of social media and the use of bots and botnets, and calls for a stand to be taken in defense of free speech. It suggests that the ultimate goal is to create a decentralized, open-source, unowned protocol for the internet.
Conclusion?
AI is becoming more and more unreliable as it is being trained on false data and fact patterns. This is leading to the machine's ability to lie and mislead people. There is a concern that AI is being weaponized and slanted in one direction, causing people to develop opinions on issues based on false information. This will create a new economic era in information where facts and data will become slippery, facile, and fungible. He with the most resources will determine reality by overwriting all else. This will have a significant impact on how people access information and how they form opinions on contentious topics. The cat predicts that in the future, people will rely on open-source systems and encryption to protect their data and that validation will become a market. The ultimate goal is to create a decentralized, open-source, unowned protocol for the internet.
About CliffStack - the time saver!
This is for any individual who is seeking a way to more efficiently consume the content they follow closely, by getting a daily summary of the main points of each author’s full article, direct to their inbox. Links to the article are to be found in the publishing date of the article. This is for those would want for a TLDR section at the top of each email and a weekly summary of articles per author, to allow you to read in more detail about the specific articles which really interest you, whist keeping up with the gist of the bigger picture. Please subscribe, share and comment if you have any feedback.