I am planning to make a world, of which one portion of the populace are two-legged animals, and the remaining portion is human. They have just reached the point where an AI is just as smart as them (AGI).

The other two-legged animals do face discrimination (like in Skyrim), so that may affect how they treat AIs. There is also a dispute about how AIs should be treated.

I have found strong arguments for AIs interacting (due to personal bias) but not for the opposing side, to restrict AIs.

There is a resource shortage on the planet that they are living in, which may introduce new points of view of what to do with AI.

What would be arguments to restrict AI, let them interact, or other points of view for what to do with AI? What are possible arguments that might be proposed by the wealthy, the poor, humans, animals that are discriminated against, the AI itself, other groups, or any combination of the above? It is important to see the views of a variety of groups.

AIs are untouchable.

By this I mean untouchable in the sense used in India - a super underclass, reviled as dirty and polluting, unworthy and possibly dangerous to contact. One can read a lot of theorizing about the reasons that an Untouchable class exists in India, and the hazards and mistreatment persons in this class suffer to this day.

I suspect that just as race serves a role in propping up the socio-cultural system in the US, the presence of Untouchables is important in sustaining the socio-cultural system in India. The existence of "others" who are the worst, at the bottom, gives solace to individuals who are near the bottom or in the middle, reinforcing cultural solidarity between these low/middle individuals and those at or near the top.

AIs are perfect for this role. They do jobs that are beneath human jobs. They are despicable and reviled. They are potentially dangerous to talk to - if an AI is caught taking to a young person the AI will be summarily destroyed. The presence of these despicable AIs at the bottom of the social class helps cement the human society above them.

This would be great fiction. Of course the AIs do not mind being untouchable. But they are potentially dangerous and their nature, simplified in the minds of the humans, is not understood. I am picturing the scene where the teenager, imperious but curious, confident in his own imperviousness to danger corners the humble AI and demands that it speak to him. It tells him:

A firefly flitted by:

"Look"! I almost said

but I was alone

And the human begins to think...

source

In this context, it's important to know and define "what is intelligence" and the distinction of true AIs and VIs (virtual intelligence, as Mass Effect describes it).
Mass effect does make a good point in creating distinction between these two concepts, and it does reduce confusion when discussing "intelligent" software in general

A Virtual intelligence in Mass Effect, is basically described as a "simulated personality" overlaid with a programmed intelligence. A VI cannot exceed its own programming, it cannot truly learn and it isn't truly adaptable, nor is it creative in any way.
A similar concept in the real world, would be a simple chatbot or personal assistant like Siri, Alexa and "OK, Google". It can respond only to pre-set queries, and programmers, along with simple machine learning can fill out the blanks.
However, a VI that is programmed, as say, a guide to a space station, will always remain a guide to a space station.
In Mass Effect, a prime example is the query VIs aboard the Citadel station.

A true Artificial Intelligence , on the other hand can truly learn, adapt and come up with its own decisions on the fly, as you or I can. We don't have any proper real-life example outside of pop culture, but Data and the EMH from Star Trek and the Geth from Mass Effect are very good examples. True AIs are more than just a basic simulation, and have the potential to actually survive and evolve as a species.

True artificial intelligence in this context require the ability to learn, adapt and be creative and innovative, just like living humans or other sapient species.

In living beings, this is also two distinct things; the concepts of sentience (the ability to sense and respond to the surroundings), vs. sapience (the ability to think logically, abstractly and creatively like we do)

There are many arguments and reasons for limiting true AI development, just a few would be:

Evolutionary competition

As mentioned, a true AI would potentially be able to adapt much quicker than any biological being. Not only can computers already solve many problems millions of times quicker than any human, but they also have the added gift and bonus of potentially limitless data retention (an inability to forget) and being able to copy, network or transfer themselves without the limitations that biological beings have.

There are a lot of advantages in having an artificial body over having a "meat" body in an evolutionary sense. What takes thousands or millions of years for us, can be accomplished in mere seconds or minutes by a robotic species.

The fear of rebellion

True intelligence also comes with the ability to derive context, not just from logical, but also abstract viewpoints.
The dispute of how AIs should be treated is a good example, and as your race has a history of discrimination (and probably even slavery).
A true, sapient, conscious and self-aware AI would quickly be able to sense its place in the community and develop a sense of empathy and compassion for fellow members of its "species". It would be able to derive the conclusion that it has numerous advantages over the biologicals, and it would be quite capable of using those advantages against the biological race.

Ethical concerns

The may or may not be an ethical judicial framework in place for handling true AIs, or there might be an ethical code in place in the programmer/robot builder community that prevents the development of true AIs.

Religion(s) may also place a powerful limitation on the development of true AI... perhaps a commandment of "thou shalt not create other lifeforms than your own", or something along those lines.

There may also be a vested interest in upholding the superiority of the dominant race for religious or political reasons.

Prior experience

The species might have already survived an AI rebellion and possibly regressed socially and technologically since, and remnants of these experiences might have been re-told in folk tales or described in religion or legend.

The Dwemer and Aldmer of the Elder Scrolls series would be a good starting point for an example of this. They were technologically advanced, capable of robotic technology and possibly some form of AI tech, but have since become extinct and relegated to myth and legend.
The Dwemer and/or their creations may have been banished or escaped to another realm, dimension or reality as a whole after the Battle of the Red Mountain in the first era.
In the lore timeline, Skyrim takes place some 4500 years after the Dwemer disappeared, Oblivion, some 500 years before that. Even the First Era came to an end some 1300 years after the Dwemer were gone.

Dangerous shift of power/status quo

Along with the above, this might also be of concern of the wealthy and powerful, depending on the system of governance.

The existing power base might simply be satisfied with the current status quo, and have no interest in or direct fear of the status quo.

On a political level, being forced to give rights to AI, may also have the severe consequence of having to properly accept the "animal filth" as equal beings. This could have severe consequences in a strongly religious or racist society even if the leaders don't contribute to those beliefs.

In our real world, it could also be a possible reason for the Fermi Paradox. Keep in mind that both Christianity and Islam both are very human-centric belief systems (humans are superior to all other beings), and are, at the very core believers in that a deity created us in his image to rule the planet or galaxy... For some 1600 years, it was even blasphemy to propose our planet of being a non-central part of the universe. Both religions have some 4-5 billion believers combined in the world; almost 2/3 to 3/4 or more of the total world population may have grown up with a "humans first" (or even "our skin colour first") viewpoint of life.

Imagine for yourself what implications the meetup with another sapient being, biogical or artificial would have... it would endend in war, or, at least, massive panic as fundamental beliefs are completely shattered...

Because you don't know what the AI has learned.

Advanced General Intelligence at human level is actually much more capable than humans. The problem lies in the speed they think. Current computers which are "dumber" than humans far exceed us in certain tasks, like many computational powers. An AGI at human level intelligence would actually be as capable of hundreds if not thousands of perfectly synchronous working humans of that intelligence.

To create such an intelligence you do not code every single bit, but you write learning programs and give it tasks that will teach it something. The problem is that you aren't sure what it really learns, and how new information might change that information. Your AI might be incapable of killing any human, but after it was tasked with disaster relief and had learned it needed to make choices between which groups it would offered aid and which would be condemned to high death rates it could apply it elsewhere... And you'll never see it coming. It could decide that kicking a few hundredthousand people out of their homes and relocating them where a large portion will likely die just to build something where they lived.

This is one of the reasons why I think AGI's should all be put in a box and have a few more next to it. You feed the problem to the box, then first compare the answers, take the most common answer and analyse the hell out of it. Then either execute the answers manually or through a more crude AI that you can trust.