AI – What could go wrong?
What is AI? The Oxford Dictionary defines AI as “the theory and development of computer systems able to perform tasks that normally require human intelligence, such as visual perception, speech recognition, decision-making, and translation between languages” This is a very basic definition. AI is always learning from very large datasets that are continuously uploaded and from its own experience, to make it better. It can write articles, create computer code, create images and videos that can be impossible to distinguish from human-created content. Think of a video of Abraham Lincoln playing Enter Sandman by Metallica on the guitar used by Eddie Van Halen! On the nefarious side, how about a presidential hopeful slapping a child, an event that never happened but looks so real? Who do you believe now? The first one is easy to spot, but the second example would be more difficult.
In 1950, Alan Turing created the “Turing Test” which says if the computer successfully tricks the questioner into thinking it’s a human, then it has passed Turing’s test. We have already crossed that threshold. Recently, Google’s AI LaMDA passed the test and even controversially convinced a Google engineer that it was “sentient.” Some AI systems have been shut down because of this.
Think of a search on Google. Without AI it gives you a list of websites that match your search criteria. For example, what started World War II? You would get a list of sites that discuss the start of WWII. Add AI and you will not get a list of sites, but an intelligent compilation of the data available to the system, written in a way that can make it impossible to distinguish from a human. It is only as powerful as the data in the model and the way you write the question. You could ask: Write an article that describes the reasons for the start of WWII, breaking it down by the top three political and economic forces and write in the perspective of a United States general in language used in 1945.
This will potentially impact many professions and industries. Here are a couple of possibilities:
- Marketing: Trigger an article to be written on your website when a partner posts a new press release summarizing the PR and pulling in additional information about the company.
- Information technology: Write code to process data coming in with the latest tools. As the tools are updated and new tools are available, continuously evaluate them and plug them into the code with new features.
Anything that can be used for good can also be used for nefarious purposes. A few examples of where it could be used against us:
- AI continuously digests publicly available data including website posts, press releases, and social media posts, to create well-written, highly personalized phishing attacks.
- AI continuously digests new vulnerability data and tests and learns about operating systems, software platforms, and applications and uses that to penetrate your systems.
AI is only as accurate as the datasets uploaded, and the algorithms written to process that data. On a global scale AI can cause significant disruption, here are some potential threats:
- Widening of our political polarization. AI can be trained to only give answers to questions that represent the most extreme views ignoring more moderate or nuanced views. We already see that on social media and search platforms.
- Significant impact on the job markets as AI can replace workers in many industries. We have gone through this many times as robots have replaced factory workers and larger more automated equipment has reduced the number of farmers etc.
- The financial markets may experience unprecedented volatility as rapid AI-driven trading creates wild swings in world markets.
All of this can be mitigated but we need to get ahead of the potential problems now.
One critical point to consider is that any data uploaded to public AI systems become part of the dataset that users have access to worldwide. Do not upload sensitive data to a public AI system. Instruct your staff who may already be playing with the new technology, not to do it either! The AI companies are providing private AI systems where you can safely upload your data and use the system to combine your private data and public data to provide private results.
What can you do now to be ready for these changes?
In the computer security field, you need to have systems that can adapt quickly to changing threats. You should already have this by using high-quality firewalls, anti-malware, limited access to critical systems, and a layered approach to protecting critical systems and data. Harden your most critical component, your staff by having regular security training and testing from a company like KnowBe4.
Start training your staff on how to use these new systems effectively and securely as soon as possible. There are many tools to help you learn this new technology such as YouTube. You can’t stop this change, but you can use it to your advantage and stay ahead of it in your job and personal life.
For further reading on this topic – refer to these articles: