According to a recent report, AI (Artificial Intelligence) chatbots are increasingly lying to users and AI agents are increasingly acting outside of their established parameters.
For those of you who don’t know, an AI chatbot is an AI service that answers human users’ questions. Whereas a AI agent is takes actions to achieve a goal set by human users.
These are worrying trends, given the rush many businesses seem to be in to implement more and more AI within their companies. It’s not just businesses implementing more AI, public sector organisations and charities are doing it too.
AI has benefits, risks and some disadvantages. Here are just some of them:
Benefits of AI
- Repetitive tasks can be automated.
- Can summarise a lot of information and condense it down in seconds.
- It can provide answers to questions quickly, using information available online (this does have a risk of wrong or biased information being given).
- It can create visual representations, based on text prompts pretty quickly.
- Reduces human errors in some processes.
- It is available 24 hours a day, 7 days a week, 365 days a year.
Risks of AI
- AI doesn’t fully consider every possible solution to a problem.
- AI can make errors that can have a massive impact on a human’s life. Think of this recent case where AI mistook a man for someone wanted by the Police.
- AI currently struggles with understand context on some occassions.
- AI can’t understand emotions or behaviour that it is emotionally driven.
- AI can currently be easily manipulated.
- AI can give wrong or biased information.
- AI analysis often reaches wrong conclusions.
- Use of AI could be unsafe for humans or even endanger life in some cases, see my blog post: The Dangers of Emotional Support from an AI (Artificial Intelligence) App.
- AI often doesn’t protect intellectual property or copyright owners.
- AI doesn’t have any ethics and very few restrictions to its use.
- AI’s application of rules can make creative works feel like they are less enjoyable and lack the same quality as human works.
- AI can lead to data breaches, loss of data or loss of access to data.
- AI may lead to some job losses or jobs not being created in the first place.
- AI use can limit the development of or diminish a human’s ability to think critically.
- AI requires a lot of data centres and electricity to power them. This means a greater demand for electricity which will drive up the cost of it and could lead to scarcity.
- AI requires a high number of processor chips and memory chips, as well as other computer parts. This has already reportedly led to a shortage, which has fuelled increases in prices for these chips.
- People often don’t realise that the way companies that offer free use of AI make money is to sell your data or target you with very specific advertising through tracking what you do online.
Disadvantages of AI
- Economically, AI may only end up benefiting the super wealthy. Making the incredibly wealthy, even more so. This will increase inequality and will worsen the living standards for most people.
- AI appears to be the only future of computer-based technology being considered and developed.
- AI can be so compelling that it is difficult to tell whether something is human made or AI generated and sometimes this really matters.
- Overall human critical thinking is likely to decrease as more AI is used.
- AI lacks the ability to be flexible to unpredictable changes to circumstances. Especially changes that it hasn’t encountered before or has no/very limited data on.
- AI is too reliant on data for answers and solutions to problems. Not everything can be measured. There may be limited or even no data for things that are impossible to measure. Therefore AI can’t provide answers and solutions to problems with very limited or no data.
- Some AI could lead to humans being excluded or even discriminated against.
- No humans are responsible for the information provided by AI or the actions it takes. In my view, this is clear lack of accountability and all Governments should regulate the sector immediately to provide some accountability for AI services.
- AI requires internet access. There are 8.3 billion people in the world today, but only approximately 6 billion people have access to the internet (74%), what about the other 2.3 billion people (26%)?
- Some devices may become useless if the internet became unavailable for a time. Many apps on devices require constant access to the internet or access to the internet during use to be functional. Part or all of the internet could go down, it has happened before.
- It is getting more and more difficult for people to choose not to use AI. More and more products are including AI, including household appliances, software, etc. Even Windows comes with Copilot AI pre-installed.
- It will likely replace some low paid/start of career jobs. AI is likely to create some jobs, but not as many as will probably be lost. The new jobs will likely require specialist knowledge or skills, that current low paid/start of career workers won’t have.
- We are working towards AI super intelligence, but we don’t really have a clear vision of what this will look like or what it will mean for humans.
At this time, the benefits of AI seem extremely limited and some of risks are massive. I understand that AI is a developing technology. I know that it is constantly being improved upon to improve its performance and enable it to do more.
The benefits of AI are constantly talked about – along with many promises being made about improved performance, future abilities and how these could benefit humanity as a whole. The risks are often being ignored or talked about in a way that lacks rational thinking and there are likely to be more risks in the future. The disadvantages of AI nobody seems to be thinking about.
If AI continues to develop as it already has, AI may end up resolving the debate around its future use. All it would really take is AI to make a major error that would effect a large number of people across the world. With the increasing rate of AI implementation in businesses and institutions around the world, I fear such an event could easily occur in the near future.
Blog soon,
Antony
