Original article was published on Deep Learning on Medium
Measures and Myths of Artificial Intelligence(AI)
Learn new about the ethics, bias, trust issues and myths of AI
Ethics are basic essentials anywhere in order to avoid the adverse outcomes of a system. AI is extremely the powerful tool can be used in ways that negatively affect the society. AI developers must ensure the ethical effects in their AI systems. They must have a realistic view of systems and their capabilities and be aware of different forms of bias potentially present in their systems. With that awareness developers can avoid unintentionally creating AI systems that a negative rather positive impact.
Some conversations of AI ethics are supposed that AI become self-aware, develop its own moral code and turn against it’s inventors. So, ethics matter more here in building AI systems.
Machine super intelligence can eventually match human intelligence, when that happens there may be risks to human society — Nick Bostrom
When developers are developing AI systems, experts must guard against introducing bias. Early AI systems are prone to bias.
For example like in Image recognition software where they are trained with certain images of living spaces like ground, gym, workspace with male and some other like kitchen, hall, bed room with female. Another example like in AI powered facial recognition systems, where trained with individuals of lighter tone skin and other with darker skin tone. This was discovered to be that trained data to the AI system was not sufficiently varied. As they featured more individuals with lighter tone skin than darker. These types of bias are extremely dangerous in real life systems.
For example: AI powered risk assessment systems, which are used in courts of some countries that help predict the probability of a person re-offended, hence provide guidelines for sentencing or granting parole, based on the calculated risk of AI recidivism. There is a concern that these systems maybe biased against people of color.
These bias can be avoided by providing effective training data and performing regular tests and audits to ensure the system is performing as expected.
In the recent days you can observe when you open any website or some Face book pages, immediately you will be welcomed to the site and asking you for queries to resolve automatically. Same thing happens when you call to some company mobile numbers. Its an ethical issue to know that whether you are speaking to a human being or to a bot.
Trust is key in developing useful, successful AI systems.
- Transparency — people should be aware when they are talking to AI systems.
- Accountability — developers should develop AI systems with algorithmic accountability, so that any unexpected results can be traced and can stop.
- Privacy — public personal information should always be protected, in the recent its has become a major issue when using third party sources.
- Lack of bias — developers should use representative training data to avoid bias and regular check for systems to avoid further.
Myths of Artificial Intelligence(AI)
- All jobs will be replaced by AI
Reality is it more create more jobs for humans, according to deeplearning.ai analysis by 2030, AI value creation is about $13 trillion in all sectors.
As per World Economic Forum analysis Automation will displace 75 million jobs but generate 133 million new ones worldwide by 2022
2. Only low-skilled and manual workers will be replaced by AI
It’s a big myth. AI-equipped robots and machinery are carrying out work generally reserved for the most highly trained and professional researchers. True, a lot of their focus has been on reducing the “drudgery” of day-to-day aspects of the work.
For example: In the legal field, AI is used to scan thousands of documents at lightning speed, drawing out the points which may be relevant in an ongoing case. In medicine, machine learning algorithms assess images such as scans and x-rays, looking for early warning signs of disease, which they are proving highly competent at spotting. human touch” procedures. These aspects of the job are less likely to be automated.
3. Companies don’t need an AI strategy
Reality is sources tell us over the next decade, there is no organization, industry, or business segment that is going to completely avoid being touched by AI. It is a risky proposition not to have an AI plan because the competition certainly will and they will be able to respond to market changes much quicker. Hard need of AI in business like A/B testing.
4. Modeling determines outcome
Surely can’t be certain of that, All AI initiatives begin as test projects. As we may get excellent results during the testing phase, but find that your model is far less accurate when we deploy it into production. That’s because AI and machine learning models must be trained on data, and that training data must be representative of the real data, or results will suffer.
5. AI eliminates humans in Earth
We mayn’t sure about it at now. As AI is in research stage not in able to its complete application. What AI we see today is basic only also AI is as smarter as we train them. It work based on trained data only. More the data more accuracy but never get full accuracy.
Is AI dangerous, will it destroy humankind?
As all know basic principle of anything on this universe has its own pros and cons. Here also applies same. Great scientists and researchers foresee the adverse effects of full development of artificial intelligence.
The development of full Artificial intelligence could spell the end of human race — Stephen Hawking
As we can see a humanoid robot like Sophia robot, is a mediocre application of AI but beyond that intelligence will may end human race.
We need to be super careful with AI, potentially more dangerous than nukes — Elon Musk
Humans should be worried about the threat posed by artificial intelligence as said before and be cautious about the development stages of AI, failed to do so will make humans to pay lives.
Kudos you did it!
I feel you have learned at least a new line of information about Artificial intelligence than before. My only intention is to share the knowledge that I have gained through online platforms. Kindly mail me suggestions to firstname.lastname@example.org. Thanks for reading.