Quick thoughts on Patrick Dixon's "How AI will change your life"
These are some of my thoughs (maybe a quick review?) after reading Patrick Dixon's book on AI.
His book is a great introduction to the topic for people who are still not familiar with what artificial intelligence is, and it shows a wide range of real world applications of AI as well as the author's predictions for the future. Patrick Dixon is a futurist and has been writing about the future of the world for many years. Since I am a programmer and particularly interested in LLMs and generative AI, I figured his book would be a good read.
At the beginning of the book, Dixon starts by explaining what artificial intelligence is and how it works. He then goes on to show us "reality checks" on how AI is already changing our lives, how it can evolve, but also outlines the risks and challenges that come with it, as well as its limitations.
Apart from his predictions, I believe that one of the best parts of this books is when Dixon talks about how AI is already being used in the real world. For example, he mentions how AI is being used in healthcare to diagnose diseases, in agriculture to monitor crops, in finance to detect fraud, and in transportation to optimize routes. I find a lot of what he presents inspiring and serves to show how much computer science can help other areas and how integrated software is in our lives.
About his predictions, I think that most of them are plausible, and well balanced, specially during the first two thirds of the book. The only issue I have with some of his predictions is that in many cases he says that "Super Smart AI" (and he uses this term a lot) will be able to do absolutely crazy things without an explanation on how or when.
For exaple, in many parts of the book he mentions that "Super smart AI" will be able to hack into government systems and go rogue if it figures that it is in its best interest to do so. My problem with these predictions is that the author implies that the more advanced AI gets, the more likely to develop consciousnes it is. Artificial intelligence is "just" statistics, linear algebra and calculus. It is a piece of software and it is not capable of "feeling like doing stuff". How and why would a computer program feel like it is in its best interest to do anything it was not meant to do? How would a computer program have will? Why would AI want to dominate the world? For humans, this question is easy to be answered. For ones and zeros, it does not matter.
Now, here are some parts of the book that I found particularly interesting and decided to outline:
- The more AI takes over, the more our communities will crave real social contact, in small groups and crowds. We see this in music festivals, football stadiums, theatres, coffee shops, restaurants, and pubs. As we discovered during COVID, ‘virtual’ is a poor substitute for hanging out with family or friends.
- AI will get better at imitating, but you can be sure that in 2050 there will still be competitive advantage for companies in many industries that (still) offer a personal touch.
- Hundreds of millions of people in hot and humid megacities are living in shacks of wood, plastic, and metal sheet, often without shade, shelter from storms, drainage, or reliable power. Many have access to smartphones and see our more affluent way of life on their screens, firing up hopes and dreams, but their own lives will not be impacted much by AI anytime soon, compared to wealthier citizens in cities of the same country.
- All leadership works by promising the future will be better in some way, and all marketing works by promising buyers that their own ‘world’ will be improved as a direct result of their own decision to purchase.
- Imagine you are walking down the street, wondering which way to turn. Post-op, you have learned to pay attention to vague notions that are hard to explain. You just have a ‘gut feel’ that the next turning is the one.
- You are wearing a bodycam and pass someone who feels intuitively familiar. You have learned that such feelings mean you may have friends in common, or work in a similar job. After some time, you find that if you really concentrate, the right word or phrase often just ‘comes to mind’.
- The future will be all about personal attention, not just being ultra-smart. When we focus on another person, a natural bond is formed. So how rewarding will tomorrow’s eighty-five-year-old find it to be listened to by a machine?
- Yet, in almost every school globally, it is still the case that exam answers are written with pen and paper. The same is true in almost all colleges and universities. Future generations will think this was completely ridiculous and negligent.
- Most teenagers in countries like the US, France, and Germany feed themselves digitally from the moment of waking until falling asleep with ideas, thoughts, opinions, innovations, news, and other information. But students often struggle to be sure what is real, which can lead to confusion and anxiety.
- One of the challenges which will come from heavy use of AI in education is the risk of creating non-thinkers: an entire generation who become used to outsourcing many of the mental tasks people do today.
- Most countries make health education a top priority. We will see digital education as equally important.
- The trouble is that generative AI simply amplifies a digest of what it has read. Therefore, AI chat responses can easily be hijacked by those promoting medical myths, dangerous nonsense, and macabre rumours as facts, if they generate enough content to influence AI.
- It will still be the case in fifteen years’ time that software developers with deep insight, knowledge, and experience of today’s coding platforms will have a competitive edge over those whose entire experience of coding since leaving university has been AI-driven.
- There is an absolute limit on the number of films the average person is ever likely to want to watch in any year. AI will create pressures on larger studios with high, fixed running costs.
- Next-generation AI search may end up destroying the quality of online content, causing it to degrade and go out of date, by removing financial rewards for website owners.
- Future generations will find it really shocking that in 2025 most digital media was still completely uncontrolled in most democratic countries, with vast proliferations of socially damaging content.
- There is no doubt that AI will play a central role in detecting and blocking unsuitable, undesirable, or illegal content, and also in detecting the age of children or teenagers who may be posing as adults.
- The same AI tools being used to clean up media content online will also be used to generate more harmful content, with ever more sophisticated methods of encryption or other ways of hiding it away.
- In future, many companies will introduce automatic bans on all social media accounts which are not in the passport name or social security number of the individual opening it, publicly displayed on all postings, with name/address/mobile phone number and other verification data to help certify the account.
- Expect a new wave of new ‘green-washing’ mis-selling claims related to environmental investments linked to false promises about capturing vast amounts of carbon, generating clean energy, and so on.
- Trust is absolutely fundamental to every business and will be even more so in future.
- Tomorrow’s websites will self-improve. Take a page which advertises a new type of vacuum cleaner. AI will create multiple ‘test’ versions of the page, and see how they actually perform in natural search (no advertising).
- AI will understand you better than you know yourself, when finding the best possible fit between how you feel personally and the leisure options in your price range.
- As soon as AI is proved statistically to be safer, we can expect insurance companies to start charging extra for every ‘human mile driven’.
- While some will prefer to have their own vehicles permanently at their disposal, at least 20% of owners of auto-driving vehicles will be delighted by 2040 to tap the ‘SHARE’ button. This will release their car to pick up other people, earning money on every hire. Such loaned vehicles will be cleaned automatically as needed, at local centres.
- All automated systems will be run by AI in future. Human rules will not be needed, because AI will analyse data for itself, make its own predictions and decisions. AI will also try to detect fakes more rapidly.
- Retail banks will increasingly find themselves in competition with financial management apps driven by AI, providing customers with better account reporting and financial planning, and an easier route to changing accounts. These apps will also make it easier to manage accounts with several banks.
- The more virtual our lives, the more personal contact matters, and this will continue to apply to major financial decisions.
- Privacy and security will remain top of the list of AI concerns among senior bankers, and this will slow implementation.
- We will soon see astonishing accuracy in healthcare predictions, linking genetics, lifestyle, and other data to what actually happens in future.
- AI will be so accurate in predicting imminent health crises that some people are likely to be denied health insurance.
- Expect debate about making it illegal for insurance companies to use such AI tools. The trouble is that these same tools will be online for people to use before buying insurance.
- State surveillance will be taken to entirely new levels by AI, with scrutiny of private lives on a scale never seen before. The primary aim of all state surveillance is to discover things going on in private lives or inside organisations that could become a threat to the state, however that is defined (including national or local leaders, damaging revelations about government agencies, etc.), enabling actions to be taken to prevent this.
- Countries most likely to use AI surveillance extensively will be larger emerging ones with autocratic governments such as China, Russia, or Saudi Arabia, and developed ones such as the US, UK, and many in the EU, Australia, and Japan, which will justify such activities as protecting democratic freedoms. Countries such as India, Brazil, Mexico, Nigeria, Kenya, Cambodia, and Thailand will use such tools less widely.
- The power of surveillance is greatest when techniques remain secret, which will mean using AI to develop new methods faster than their discovery. The need to keep techniques secret will often mean that police cannot prosecute, because the risk of disclosure is too great. This will be a huge issue: greater surveillance insights from AI, plus greater reluctance by security services to reveal the extent of surveillance. The result will be pressure to bend rules, invent evidence, and circumvent legal processes, to ‘take people off the streets’ who are considered a risk to the state, without telling the truth as to why.
- You could find yourself unable to interact online with a group of friends that AI objects to, that it thinks are unhelpful/disruptive influences – calls and digital links blocked. Use of encryption-based social and digital links blocked. Use of encryption-based social media or internet searches, or VPNs, could result in fines or worse. Even if encryption can’t be broken by AI, activity will be detected.
- AI will be used to deliver a vast range of messages, often subtle, tailored to each person, and so the ultimate propaganda tools will be born. For example, we know that text or images flashed for a fraction of a second in front of someone’s eyes can sometimes change how someone thinks or feels, without them realising.
- Expect growing debates about who is a patriot, freedom fighter, terrorist, or criminal.
- By 2030, AI will also be used for assassinations, using drones with facial recognition capabilities. Such drones will only need a second or two at low height to confirm identity before attack. With sharp blades, they will cause fatal injuries to the head, without an explosive charge that could kill others nearby.
- Perhaps the greatest AI danger we face in the future is our own AI systems being deliberately turned against us as a weapon. Not just one form of ordinary AI, but multiple variants of Super-Smart AI.
- At times of great tension, such fakes could help trigger conflict. Hostile states will create all kinds of convincing fake news – for example videos of police firing on unarmed protestors. All to create a fog of disinformation, leading to riots or worse. AI will be at the heart of such deceptions.
- A security officer can access all of someone’s personal information relating to internet activity (searches, social media, purchases), location, phone logs, family and friends, employment, tax, immigration, banking, driving offences, criminal record, health records – plus other insights generated by AI-linked pattern recognition, predictive analytics. Such an entity will become a primary target for hostile countries to attack (and for ‘friendly’ countries to quietly probe).
- This weakening will be accelerated by AI, to the point where democracy is effectively killed off in several more countries around the world by 2060.
Although having criticized the book, I realy enjoyed my time reading it and think it is worth it for anyone interested in the topic.